Interview noteThis page restates a public interview in new words. Dollar results, win rates, and payouts are claims made by the speaker or the host. They are not this desk's record, and they are not a result to expect.
Fabio Valentini is the scalper the host introduces as a multiple top-five finisher in the World Cup the caption calls the Robins Cup, and as an alpha researcher for more than one seven-figure firm. The host is Brandon. A large part of what Valentini does is already public. Asked what he would still teach a competent trader in person, he starts with data analysis and with risk protection. The struggle of his career was a process that can tell an actual edge from a hope, and a way to manage the risk once the edge is named. The hour is that process, and then one model on a Friday Nasdaq chart. Nothing in it is advice.
Stop when the data says the edge is gone
People widen a stop until it is larger than the target. They close early because hope says the market will reverse and hand the gain back. They do not know when a strategy is finished. He uses the opening-range breakout with volume. It is used worldwide. It can be tested. There is research, he says, on a drift in the Nasdaq. That is an edge. The question is 2028, if a lot of people are using it and, net of commission, it is no longer profitable. Validation is what says to stop. Competence is everything that gets a trader to that decision before the model is turned on. You execute when you know the thing being executed works.
The other half is automatic. One function on Deep Charts blocks the trader when the day's budget is spent. A trend-following strategy on a day that is consolidating and going nowhere, with no daily budget, keeps taking the same setups. Three days of that consolidation, and he says 20 or 30 percent of the account can be gone. When the budget is met, he survives the day and comes back. He thinks that is one of the larger lessons for someone starting and for a professional. Separate the idea from the day. Years of trading, and capital managed for yourself or for someone else, do not make a person immune. Follow the plan. Lose the money. Done. Train, or a sauna. Do not spend another three hours at the chart trying to recover. It gets worse.
He was fully discretionary. With the tools available now, including what is coming in AI, spending days backtesting is not efficient. Validate at least the individual trigger with statistics. He uses three phases. The first is the bias. Intraday, multi-day, and long-term, he needs to know where he is going. On Bitcoin the practical version is on-chain data, fundamentals read off a public blockchain. Portfolios holding more than a thousand bitcoin, new portfolios he reads as retail, and what BlackRock is holding, are all open. He pulls that into one proprietary model and builds the bias from there. In the example, they are supposed to be bottoming and the halving is affecting the price, so the bias is long.
The trigger is when to execute. It can be automated. Absorption against demand. Demand, for a trend, against an order block. An order block against the value-area low. Swing trading, twenty or thirty or forty opportunities a year, he does by hand. One weekend, a checklist on the order flow. Short-term order flow is hundreds of thousands of setups. Automate the trigger, isolate it, and watch a one-to-one reward. Of four triggers, which one performs. The most efficient short-term trigger, aligned with the long-term bias, is a model. The problem is that the shorter the horizon, the more easily the edge breaks.
He was stubborn in 2025. A model that had performed, in his words, amazingly in the World Cup he traded, started to fail after Trump. The new-year sessions were more compressed. Explosion, then back inside the range, not explosion and continuation. A pure trend-following momentum model underperformed. His own analysis said the payoff ratio was deteriorating and to stop. He had used the model for multiple years. He told himself another month, then another two. He thinks he could have saved probably six figures. Those are the errors that produce a trader who can adapt. A trader and a machine, now, have to work the same way.
A sideways equity line, then more than one model
Edge decay has a signature that is not a single losing streak. He wants an out-of-sample period that is not respecting the in-sample results. The equity line that was floating up starts to go sideways before it goes down. It does not drop immediately. Sideways, net of commission, is break-even, or a slight loss, or gain at a much slower pace. He evaluates it on a rolling quarter, the same way he did with the World Cup. The strategy is better in some stretches and worse in others. For scalping, three months gives him more than a hundred executions, enough to compare something like two hundred with the previous quarter. Three or four swing executions can be a negative streak and still be noise. Two hundred or three hundred executions that have flipped down, and he stops immediately. Another quarter eats the gain of the quarter before.
He does not trade one model. In 2023, for more than six months, a momentum model was in a drawdown because the market compressed again. He will not put a number on the rest. He made money in crypto with the on-chain model, with intraday models, and with global macro. The secret, as he frames it, is not depending on a single strategy, because every strategy meets a market that does not match the idea. The problem he assigns to most traders is expectation. Gain in any condition. A 100 percent win rate. One strategy for life. He has changed countless strategies. Zero-days-to-expiration options started to account for 60 percent of the notional volume per day, and the volume was doubling year by year. An intraday trader cannot ignore the hedging and the option flow. He did not want to study options. It is not easy for him. It became necessary. Even then, he cannot promise next month. The distribution of returns is not linear. It is not a salary. Two negative months, one month that covers more than the prior six, two months at break-even, one incredible month. During the years the secret became validated edges on multiple markets, diversified so the result is smoother. The host says the only holy grail is diversification. Valentini agrees.
The numbers he wants people to expect are the opposite of what he calls the propaganda. One of his quantitative systems, which the caption renders as IVB and also as IBB, has a negative reward-to-risk. The common instruction is one to twenty, one to thirty, one to forty. He says you can make a gain with that if you are good, and you will also have streaks of ten or twenty losses. A model needs a payoff ratio, a recovery rate, and an equity line that is smooth. He will not choose, between two strategies, the one that swings wildly and arrives at ten million. He will choose the one that builds consistently and arrives at two million, because it lets him build more capital while expecting less drawdown. The first is a lottery. Enter during the downswing and that account is probably in a margin call. He is against building the whole portfolio that way. He is not against risk. In a bull run he put his foot on the gas and loaded heavily, because he was convinced of the analysis. That is not how he wants an entire portfolio built.
A payoff ratio of 1.5 or 1.6, across thousands of trades, he calls absolutely amazing. A single week of his scalping can print a payoff ratio above three, on twenty executions. Stretch that to a hundred trades, or to a thousand, and a sustainable payoff ratio of four, five, or six is probably high-frequency trading or overfitting. Vibe-coded strategies, he says, are full of look-ahead bias. They look perfect and then fail live. Done properly, a 1.5 is amazing, and the IVB at 1.44 over five years of data is amazing to him. It will not make life-changing money on one trade. The equity, in Sharpe ratio, beating the S&P 500 is the goal he names.
Measure the event, or you are not measuring the same thing
Discretionary traders produce the contradiction. He knows remarkable ones who pull several data sets at once. A strong gamma level, an absorption, order flow, fundamentals, and options. Telling that person to backtest every confluence runs into the data. Short-term depth, on a one-minute chart, is difficult and expensive. The CBOE feed, he says, is $30,000 a month. Most retail accounts cannot pay it. He still thinks the experience of a genuinely experienced trader beats the algorithmic method.
How that trader improves, when the problem is not psychology and not risk, depends on the strategy. One swing model he was also using in a live session at BNP Paribas is the earnings surprise. Expectations are neutral. The earnings beat them by 50 percent. The event is measurable. Everyone sees the same print. A large buy aggression creates slippage, so the large participants will not buy the whole position at once. With what he calls a really good statistical validation, the market returns to the gap made by the surprise and then goes back up. The reward is asymmetric because the cover is below the gap. A stock, if the name is chosen well, usually goes up. He says you can take the last five years, see expectations beaten with a 70 percent probability, and a drift in the sixty days afterward 80 percent of the time. That is an edge. A trend-line breakout is not. His line is not the host's line and not a machine's line. There is nothing to improve if the object being watched is not the same object. He allows discretion because he does analysis. Order flow is different. Five hundred contracts executed at a Nasdaq price are visible to him and readable by a machine given the data. The discretionary problem, for him, is the lack of an objective data set. An engulfing candle is the same trouble. One trader wants the candle above for a week. Another wants the body above. An educator can teach a strategy that the student interprets into something else, and the student is lost. What he says in public is not to trust him and not to trust the next person. Take a data set. If the trading is discretionary, observe, find a pattern that repeats, understand why it happened, and execute the edge in simulation. Do not risk real money. That was his own error. He started in binary options, which he says are now banned in Europe, and he lost a lot of money trying to guess the protocol for executing something that could have been measured.
The same measurement is not only for an algorithmic trader. An order-flow trader who can measure a trigger can improve the entry, and the entry changes the equity line. An intraday example: move the stop to break-even at one-to-one, because the market either drifts or comes back, and save twenty or thirty stop-losses a year. For a strategy with one or two trades a day, that is a large difference.
He will not call a signal bullish until the outcome has been tested. He walks in as a spectator. He does not bring a belief. If the data say the trigger has a better rate used the opposite way, he uses it the opposite way. Absorption is a lot of pressure that cannot break a level. If the level does break, there is an acceleration. Do not use it to time a bottom. Put a stop order below, wait for the break, and let it accelerate. Nobody has a crystal ball. The data say the same thing to him, to Brandon, to Andrea, and to anyone else. He tests signals that ordinary trading says should mean the reverse. An RSI at an extreme is supposed to reverse. The index was built as relative strength. Down there, it is telling you there is a lot of selling. Catching it is catching a falling knife, and that is what he says multiple models validated. On the Nasdaq version of that system, his first assumption was that a break of the start of the range, a strong candle, would bring the larger expansion. What the test showed was a small candle with a lot of delta. A big candle with a lot of delta is usually where the market maker rebalances. You cannot cover immediately. You get taken out. Starting from "I think" and "I hope" has no way through.
Before the test, he wants the reason the edge exists. Stocks have an upside skew. That is why the IVB exists and why it persists. Commitment-of-traders data force large participants to disclose positioning. That is why a drift exists. An earnings surprise forces large players to avoid slippage and to build a cluster of limits in the direction of the stock. Politicians have an informational edge. If he is asked why he uses something, he wants to be able to walk it from A to B.
A market maker he calls Mati, after Valentini's own video on large trades, ran three years of data and found what Valentini had found from experience. A large print that is one operator, someone opening five hundred Nasdaq contracts, is probably a nine-figure account. From a statistical point of view that order influences the market five, ten, and fifteen minutes later. That is the window Valentini uses to get in and get out. It is a directional bias, and it is not the only input. In two years the edge can stop existing. He can still explain why it existed. Someone he calls Freddy, with experience inside a market-making firm, can hand a quant a question: gamma exposure, strong absorption, a threshold, and then the adverse excursion and the expansion. If every time price reaches a gamma level in that condition the average excursion is about twenty points and the directional excursion is about sixty, the average reward is known and the edge is validated. Then be careful about how much of the edge is described. Retail, he says, does not have large capital. The more hedge funds execute the same hedge, the more the gain of the edge is pulled toward zero. They close the inefficiency.
Add one piece, and throw it out if the metrics do not move
Confirmation, and the fear that it only gets you in later, is handled as edge contribution. Bias first, because a position needs a direction. He mainly trend-follows. Then a new piece of information goes into the model, and the question is whether the signals get better. He is not a quant. Other people run the process. Absorption, aggression, exhaustion, switched on and off, for that bias and that asset. Then the choice. A fully automatic system needs a platform that can permute parameters and run Monte Carlo. A discretionary edge, which is how he uses one of his, means he is the analyst, the person who executes, and the risk team at the same time. The daily plan has a budget. Five thousand dollars today, a thousand per trade, five executions at most. Worst case, minus five thousand. Best case, he states twenty-five thousand. The bias can be option flow or the structure of the volume profile. The trigger is whichever of the tests won. The area, the direction, and the trigger are set before the session. He has a budget, a model, a statistical skew, and an explanation of why a long resumes at a put wall and what the market maker is doing there. The point of owning the model is that when it goes badly, and it will, he knows what to change. Justifying the model to himself is how a trader gets lost. Even a fund he names, Medallion, with quantitative models, has had good years, average years, and amazing years. The gain and loss floats. When the emotion arrives, the reply has to be specific. Absorption comes out of the execution because the last three months were poor, or because the market keeps leaving him unfilled, or because the fill is good and the reward is worse. That testing is what people do not want to spend time on. They want a package. He has given away, for free, how he uses order flow, option flow, on-chain data, and swings. The request that remains is the strategy that works forever. He can give everything he knows. The market is a changing creature. The tape under Trump is completely different from two or three years earlier. COVID was completely different from 2010. An algorithm written in 2010 that has lost money for ten years, and is still being executed because it worked in 2010, is what some retail traders are doing. The strategy is not paying them. They like it. It makes sense. Education taught them that people prefer a simple explanation. Take this demand. If trading were a line on a chart, one data set, and five minutes from a sofa, the way the information-product business sells it, hedge funds would not be paying billions a year for research.
A model that is amazing at a 300 percent imbalance and falls apart at 301 or at 299 is overfitting. He describes a three-dimensional chart of profitability against the parameters. A sharp peak means the system works at one setting. A real edge is a flat surface. An imbalance from 250 to 300 works. Four hundred is too much. Two hundred is too little. That is a confidence interval. Three hundred working and 301 destroying it is the correct parameter for one asset, and he sees it constantly in strategies people are coding in public. An equity curve, a Sharpe ratio someone will call seven, a payoff ratio of 24. A strategy with that expectancy, on a one-minute or a fifteen-second chart, could flip accounts fast enough to build generational wealth in a short time. Then commission and slippage are added and the edge is gone. Sometimes the test simply forgot them and assumed a perfect fill. One tick of slippage against a strategy that makes one tick, with zero commission and zero slippage, can be made to look extraordinary. Coding has opened testing to everyone. The missing piece is a procedure. Without it, the test and the market are different things.
A long thesis that fails, and the short of the same idea that works, is not an automatic reversal of the model. A strategy can lose money consistently to commission. He keeps specific live accounts for testing. The commission is large. It is a cost he allows so that he can later run size at a hedge fund. A platform will not reproduce the short-term execution, the slippage included. A strategy that steadily loses does not, flipped over, become a strategy that steadily wins. Commission sits outside the idea. He will abandon the strategy even when he believes the thesis, if he cannot find a consistent edge. Executing because the thesis feels right, while the edge is eating capital, is the ego. A portfolio can hide the same problem. Six models, four of them excellent, two of them slowly losing, and the portfolio still rises or looks like break-even. Hundreds of executions say the edge is decaying. That is the proof the strategy has to change.
The opening question, how the same order-flow data can still be an edge for everyone, gets a no. An edge does not work for everyone, because an edge is a composite. Order flow is a data set. Option flow is a data set. The commitment of traders is a data set. The data set is not the edge. Option flow, with Freddy, is the CBOE feed at $30,000 a month. Without the ability to process it and interpret it, it is useless. He compares it to the parts of a Ferrari spread across a room. Infinite signals can be built from one data set. If order flow meant only buy the aggression and sell the absorption, it would be useless. A single minute of an order-flow candle contains enough to process market-by-order, bid and ask, and a delta profile. Raw material. Then the bias, the option flow, and the profile. Three or four steps. Fully quantitative order-flow edges exist, he says, and they live on the extremely short timeframe. Those can decay. Hand one to a firm he names, Jane Street, and the capital closes it. They squeeze what is there, and the edge is finished.
Friday's cage, and a one-to-one that is allowed to reload
The chart is a Friday that consolidated. The problem on that kind of day is the relevant swing. People do not know when an impulse is finished, or when the second drive can be bought. The template, from the team at the platform, marks who is dominating. From the open, the first drive prints a red cluster where he had already seen strong absorption to the downside. Buyers pushed, a triangle on the tool, and were absorbed. The area printed. That is the first place he would engage, and he does not yet have a swing that starts there. The session opened with sellers absorbing, price came down, and buyers absorbed. He calls that the cage. The profile from the start of the session is balanced. Buyers and sellers are at fair value. Neither side has control. He wants the directional auction outside value, because that is where a directional move can exist.
The first setup is VWAP, the volume-weighted average, where a reload shows the maximum pressure to the downside. The area collapses. The sellers in that print are stronger than the players they ran through. Sellers took control. The next area is lower, and the distance from the first point to the second is the room for the trade. At the bottom, buyers take an important aggression upward. The average is engaging price on the downside, and the delta colors say the buyers are dominating the auction. Price returns to value. At the value-area low there is another confluence, a line where the buyers took complete control. Sellers try to bring it back and fail. That is an absorption.
Delta, for him, is the main force. A reversal from the open wants strong aggression on his side. Sellers in control, or buyers punching a wall. The buy aggression outlined at the top was absorbed. It added no reward. He calls that a mean-reverting setup, and the same picture appears on a continuation. Buyers are aggressive at the top, the result is zero, and the candle can be negative. The buy power was absorbed to the downside. The delta candle is also how he knows when to put the risk at zero. He enters, the stop is placed, and when the sellers take control the risk comes off. The other setup is the buyers reloading. Seller pressure, a candle with no result, fully absorbed, and the buyers taking control. That one was impossible at the bottom, because it was the bottom. The retest of it is possible. The sellers, back in the area, make no result.
Nasdaq effort is the proprietary study of who is actually dominating. It simplifies the side in control by looking at imbalance, absorption, aggression, and the path of least resistance. If a given number of contracts moves the price up more easily, the book is thinner above, and the market can explode there. Those are the confirmations. Sellers take control from a box, he has a timing area, and he could have tried an explosive position back toward the next zone. Returning to an area that already showed absorption, he sees a stack of two seller-control zones. Extra confluence. On the long, he skips the first because it was too fast. The second gives him the timing at the close of the candle. The candle says it is time to go up. He trails under the last aggression. He expects the market to continue, to dip, and to resume. If he wants back in, he reloads when price is inside the area again. The trail takes him out. A separate tool marks the swing. The market-by-order signature can be set more than one way. Three hundred and forty-five orders might be one participant or many. In a long, if buyers are aggressive at a level and price cannot advance, he may trail under the last aggression instead of waiting. If that aggression fails, the market reverses. On the other side, buyers pushing the auction with three hundred contracts are confirming it, and that is when he trails. Inside the body of the candle he will cover below the print, because he expects that participant to reload there. A hundred and eleven contracts absorbed at the top are just a data set until a bias and an idea validate them. The applications, algorithmic and discretionary, are numerous. Buying a feed and expecting to be profitable is the same, he says, as buying a data set and expecting to be an analyst. You have to be the analyst first. Where the market followed the aggression and then the buyers, on the return, tried to push higher and were absorbed again, that is what he uses. A print he reads at 167 contracts is a lot. Buyers then pushing the other side is, usually, how he sees a market bottom. Confirmation from the opposite side.
The profile can be read with the price removed, from the bottom swing to the top. Balance is what he calls the land of nowhere. Outside it is a discount. The image is an Apple store and the price of a new iPhone. Below that price, you are discounted, and he always wants the area marked. Buyer aggression that is actually relevant, a lot of sellers absorbed, and the value-area low. Tested properly, price fills the level. The orders were executed. Anyone can see them. The aggression from that zone was more relevant than everything that happened in the balanced area. A buy wants its confirmation, and the confirmation should be used to buy from a level the order flow already cares about. This version of the model is a scalp into the intraday. It takes short-term positions to catch the drift of the day. Aggressive scalping, in his distinction, is not that main drift.
What he calls the order track watches, inside each candle, who is dominating the migration of the value area. Sellers dominate one candle, a large delta, a profile that is fairly balanced. The next candle has a lot of sell aggression, closes the other way, and the value area is still lower. The switch is confirmed when the value area itself closes higher. They were in a downward drift. Value was shifting lower. Now it is shifting higher. The candle where the sellers dominated gives way to one where the buyers do. That is how he loads a one-to-one. It is in the direction of the order flow, and the reward is consistent. If the main direction of the day is understood, every movement in that direction can be scalped. Three or four positions at one-to-one, instead of one position at one-to-five. A consolidating market can still pay. A one-to-five can sit in a single position for a long time. If the idea is validated and value keeps migrating higher, with the candle still green, he expects the value area to protect the position and to reload higher. The entry is at the value area, the cover is below it, and the target is another one-to-one. The same at the close of the next candle. One of those one-to-ones can go against him and take the stop before it works. There are still compressions where the target is hit. On the sell side, value that had been switching higher starts switching lower and the sellers dominate. The stop is above the protected value area, and he takes one-to-ones as the market continues down. A candle closes, he takes the target, and he waits for the market to realign. A sell candle that is still closed inside the area is not the next short. A break below would be the same position. If the market says no and flips from the value area up to the value-area high, the stop goes below the value-area low and he takes another one-to-one. A switch lower on a green candle is a pass. He waits for the red, and then the switch lower is another position.
Inside the candle, discretion is allowed. The point of control, lower than the previous candle, means value is migrating down. The next candle tries to break that level. A lot of buying, and zero result, means passive orders are still there. Seeing that absorption, he can get a little more aggressive and squeeze a one-to-two. The objective of the model is not a huge reward. It is a consistent equity curve, and then, maybe, risking the gain already made on a larger position. The win rate is not constant. A drift can hand three or four targets in a row. He walks one sequence: a switch lower, a target, a second position unfilled, a second target, a switch long and a target, a stop, and then the market down with two more targets. The drift can be positive. It is not evenly distributed. The win rate of this model, he says, is really high. He will not automate it. The automation does not include the refinement of the position. The model would leave the risk where the candle closed. He sees a large buyer absorption, trails, and takes the risk to zero. When a drift is sustained, the way the closing range was, he builds. A flip higher, and he can be at the absorption or just beyond it, stop underneath. The market confirms. The first position goes to breakeven because he can cover under that aggression. He loads another at the extreme delta. On a day with a real drift he can have three or four rolling positions, all on the initial risk of one. The reward is not linear. It is exponential. Extremely choppy days are the negative ones. The win rate can sit around 40 to 50 percent, which at one-to-one, plus commission, is break-even. It is not a holy grail.
He has only taken the time to test it on the Nasdaq. The migration of value, and the actual orders, might work elsewhere. He will not say so. He has not done the testing, and he does not want to spread that as information. He has models that make money on the Nasdaq and lose on the ES, and the reverse. Correlation is not the same microstructure. A strategy that is profitable in gold, carried onto the Nasdaq because the two sometimes move together, is an assumption he calls madness. It is the same assumption as expecting to trade well because someone else has.
The model is six months old. Before that it used aggression and the large trades, and not the value area. The value area is what tells him how sustained the drift is. The longer it keeps drifting up, and the more it builds, the more position he can build. A large order can be spread across candles. The next candle retraces, the large order is visible, and a trader closes. The value area is what keeps him in the drift for as long as the drift lasts. He also gives gain back in the choppy sections. At the moment he only trades trend-following models. When the regime is wrong and the market is not moving, giving some of it back is part of the game. Asked what matters most, he says the value area together with the delta profile, and the last piece he added is the Nasdaq-effort zone, the same one he uses in another model. Value migrating higher, and that zone saying the pressure is skewed to the upside, and every input is pointing the same way. Filtered for every confirmation, the chart he walks is six trades, five targets, and one stop. Migrating lower, he would only take the targets that meet the same filter. He is wary of overfitting. Stress the data long enough and it will say what you want. He has been an algorithmic trader. In real time, the way price reacts to the area is how he sees the order flow influencing the next short-term drift. A choppy stretch on the same tool is a precise short, another precise short, a long that stops, and then three stops in a row. That is the market developing. Every trend-following system suffers in a consolidation. He has accepted it. He does not blame himself for a consolidation week. The goal is to protect the capital. He compares it to running tourism in the UAE and accepting that the summer, about two months a year, does not bring many tourists. Commission is the same kind of cost. Gain with commission requires an edge that clears the commission. The short-term impact is larger. A model that keeps the trader on the correct side of the market can do it.
He does not want the model used with the brain switched off. Test it. Use simulation money. Do the research. Execute only when the conviction that it is an edge is complete. He is spending heavily on quantitative research, on bias systems that also take in option-flow pressure and hedging activity. He thinks the model can be improved, and it will take more time. Validation on this one used at least three years of data, and then three months out of sample, because the sample is large. Three months can collect maybe three hundred trades. He has not published it as a standalone strategy. It is a study and a set of indicators. The intended use is confluence with a model the trader already has. The error is making the indicator the authority and stopping the work of deciding which way the market is going.
Someone who understood all of it and still struggled is, in his view, meeting the fact that any model, in one slice of time, can be profitable or negative. Edges are long-term. The advice is not to rely on one model. He also runs a longer-term mean-reverting trend model. Some days the short-term model is quiet and the long one is not. Some days both are quiet and the crypto model is not. He has not found, in a lot of testing, a model that is profitable in every environment. On a prop account the constraint matters more, because going underwater can burn the account. Diversify, or at least have a way to stand aside when the condition is absent. The empty stretches on his chart are the filter doing that. If it could delete every consolidation and keep only the directional moves, he says he would be a billionaire, and it would not be a system. It would be a printer.
He keeps both the early entry and the later one. One is the daily shot, the area where he is confident in the long. The other is momentum. If the area breaks and the tool keeps printing, the momentum model says he can be in the first place or he can be further along, with the pressure. A positive reward and a win rate that holds are enough. He would not skip it. The error is demanding the whole drift from a target placed at the top, when the asset is already overpriced. The market retraces, the position is closed in fear, and the rest of the errors follow.
For this model the management is a trail, not a partial. Two Nasdaq-effort boxes. He enters at the first green box. When the second prints, the stop goes to breakeven. Another box, and the stop covers under the second. Another, and there is a cost to the trail. It walks him out. He tested partials. On the Nasdaq, which can expand in one direction all day, a partial can take him out and leave two hundred points of rally behind. The days with a real rally pay for the partials that were not taken. On currencies he does take a lot of partials, because that market mean-reverts more. The Nasdaq has an upward skew. On the algorithmic side the partials hurt the gains more than they reduced the risk. When the trade is right, they eat gain. When it is wrong, the full stop is taken anyway. The risk comes down a little. The gains come down more. It has not worked for him.