Your trading habits are a second strategy, and nobody has tested it

Most traders who lose money assume the problem is the strategy. So they go looking for a better one: a new indicator, a new course, a new Telegram channel. Sometimes that is the right diagnosis. But every trader who runs a strategy by hand is really running two strategies at once: the one they wrote down, and the one made of everything they do differently from it. The first one usually gets tested. The second one almost never does.

Losing is the norm, not the exception. SEBI's study of India's equity derivatives market (press release of 23 September 2024) found that 93% of individual F&O traders lost money over the three years from FY22 to FY24, about ₹2 lakh each on average, including transaction costs. Data like that cannot tell us how many of those traders had a sound plan and executed it badly. A simulation can at least show what execution habits are capable of, so we built one in which the strategy is known to be good and only the trader changes.

The strategy: modest, profitable, and boring

We used a hypothetical trend-following profile: it is wrong more often than it is right, but its winners are twice the size of its losers.

• Win rate: 40%. Six trades in ten lose.
• Payoff: winners make 2R, losers lose 1R. ("R" is the amount risked on a trade: the distance from entry to stop, times the position size. If you risk ₹5,000 per trade, 1R is ₹5,000.)
• Risk: 1% of the account per trade.
• Costs: 0.08% of the account on every round trip, standing in for brokerage, taxes and slippage.
• Frequency: 150 signals a year, for three years: 450 signals in total.

Here is why a strategy that loses 60% of the time makes money. Over ten trades, on average, four winners bring in 8R and six losers give back 6R. Costs take about 0.8R. That leaves +1.2R per ten trades, or +0.12R per trade: about an eighth of the amount risked. Real edges, where they exist, are often this small. Hold on to that number, because everything that follows is about what happens to it.

Four habits, modelled honestly

We simulated 10,000 traders running these rules exactly as written: every signal taken, every exit honoured. That is the benchmark of perfect execution, which automation approximates but never fully achieves. Then we gave the same traders the same signals with four habits switched on, first one at a time and then together:

The habits only act on what the trader could see at the time. To make that possible, every trade follows a simple path: it first reaches +1R or first hits its stop, and from +1R it either runs on to the target or falls back.

The question that decides everything: what happens next?

Whether booking early or moving a stop costs money depends entirely on what prices tend to do after the trader acts. So we ran every scenario in three market models. All three give exactly the same result for the rules as written, 40% winners. They differ only in the odds that a move continues, compared with a fair coin between the same price levels:

• Trending: moves tend to continue (continuation odds doubled). A trade at +1R is more likely to reach its target, and a trade that has hit its stop is more likely to keep falling. This is the market a trend-following strategy is designed for: its edge comes from winners that keep running.
• Random after entry: the entry has an edge (60% of trades reach +1R before the stop), but from there on, price moves like a fair coin.
• Snapping back: moves tend to reverse (continuation odds halved). Most trades reach +1R, but many fall back before the target, and stopped-out trades often recover.

Real-world evidence suggests that, over some horizons, markets behave more like the first case. Terrance Odean's 1998 study of 10,000 US brokerage accounts found investors sold winning stocks roughly 50% more readily than losing ones (a pattern called the disposition effect), and the winners they sold went on to beat the losers they kept by about 3.4 percentage points over the following year. That is over months rather than single trades, but the direction is the same.

In a trending market: the "safe" habits are the expensive ones

Here is the market this strategy is built for. Each row is the same 10,000 traders and the same 450 signals.

"Median" is the middle trader's three-year return; "unlucky 5%" and "lucky 5%" show the spread, with one trader in twenty doing worse than the first figure and one in twenty better than the last. Edge per signal is the average result of every signal, in R, after costs.

Same market, same signals, same order. The habits turned a strategy where 19 traders in 20 made money into one where 4 in 5 lost, with the typical account down 22% after three years. Even the rule-followers had a rough ride (a median worst drawdown of 19.1%, and 5.5% still finished behind), but with all four habits the typical worst drawdown nearly doubled, to 36.8%.

The two habits that feel most prudent did the most damage. A strategy that wins 40% of the time lives on its winners: they are scarce, and they must pay for six losers in every ten trades. In a trending market, a trade sitting at +1R is usually on its way to +2R, so closing it "to be safe" gives up most of the run. The moved stop does the mirror-image damage: a trade that has hit its stop is more likely to keep falling than to bounce. Yet in our model more than four moved stops in ten still recovered to breakeven, and that is the trade the trader remembers. A habit that pays off often enough to feel right is far harder to break than one that never does.

The fair test: other markets

We promised to publish the results that don't flatter our argument, so here is the same experiment in the other two markets.

When prices move randomly after entry, booking early and moving stops cost nothing on average: in a fair-coin market, no exit rule can add or remove value; it only changes the shape of the results. When prices snap back, the same two habits help, a lot. Booking early more than doubled the edge.

Two habits behaved the same everywhere. Missed signals and impulse trades cost money in all three markets: missing a fifth of the signals gives up a fifth of the edge, and trades with no edge only add costs. They are pure execution leaks.

The real lesson: discretion is a strategy, so test it

If booking early can destroy one strategy and rescue another, the habit is not good or bad in itself. It is an exit rule, applied by feel, that has never been tested. In a snapping-back market, a trader who books early is not showing judgement; they are accidentally running a better strategy than the one they wrote down, one with a 1R target. The right response is not to keep improvising. It is to write that rule down and test it, so you know whether it works, in which markets, and by how much.

That gives every habit a clear fate. Discretion that genuinely helps becomes part of the rules, where it can be tested and applied consistently. Discretion that hurts gets removed. And the pure leaks, the missed signals and impulse trades, get removed regardless, because they cost money whatever the market does. You cannot tell which habit is which by feel, because each one feels right at the moment you act on it.

We also halved every habit (10% of signals missed, 20% of trades booked early, 15% of stops moved, one impulse trade per eight signals) to check that the trending-market result was not just harsh assumptions. Occasional, mild lapses still cut the edge from +0.120R to +0.029R, the median three-year return from +63.5% to +9.0%, and raised the share of traders losing money from 5.5% to 39.0%. When the edge is an eighth of the risk per trade, there is very little room for untested improvisation.

The case for rules, and its limits

None of this means traders are foolish. Each habit is a reasonable response to discomfort. Booking a profit relieves the fear of giving it back. Moving a stop postpones the pain of admitting a mistake. An impulse trade relieves boredom, or the itch to win something back. The trouble is that those feelings are strongest exactly when the rules matter most.

This is the real case for writing rules down precisely and letting a machine execute them. (We sell that service, so weigh the argument on the evidence, not on our word.) Automation does not make a strategy smarter. It makes it consistent: it takes every signal, applies exactly the exits that were tested, and never trades because it is bored. Whatever judgement you have goes into the rules, where it can be measured.

How to test your own second strategy

You do not need a simulation to measure this in your own trading. You need a written rule and a spreadsheet.

1. Write the rule completely. Entry, stop, target, position size, and what happens to a trade still open at the end of the day. If any part says "depending on how it looks," it is not a rule yet.
2. Keep a shadow log. For every signal, record what the rule would have done and what you actually did, including the signals you skipped.
3. Compare the two in R, not rupees. After 50 to 100 signals, total the R of both columns. The difference is the result of your second strategy. Over so few signals that difference is noisy, so treat it as a hypothesis to test, not a verdict.
4. If your discretion wins, make it a rule. If you keep booking early and it keeps helping, test a closer target as a written rule on the strategy's full history. If it holds up, adopt it, and stop doing it by feel.
5. Pre-commit your exits. Place the stop and target as orders the moment you enter. Many brokers support this, for example as a GTT/OCO ("one cancels the other") order that holds both exits at once, so the decision is made while you are calm.
6. If the gap persists, take yourself out of the execution. Simplify the rule until you can follow it, or have it executed automatically and keep your role to design, monitoring and review.

The goal is not to stop thinking. It is to do your thinking in advance, when the rules are designed and tested, instead of in the middle of a trade, when every instinct is pulling the other way.

Methodology

Traders per scenario

10,000; identical random draws in every scenario

Signals per trader

450 (150 a year, 3 years)

Strategy

40% reach +2R, 60% stopped at −1R

Trade path

+1R or stop first; from +1R, target or stop

Trending

50% reach +1R; 80% of those reach +2R; moved stops recover 3 in 7

Random after entry

60% reach +1R; 2 in 3 reach +2R; moved stops recover 6 in 10

Snapping back

80% reach +1R; 1 in 2 reach +2R; moved stops recover 3 in 4

Risk per trade

1% of current equity, compounded

Costs

0.08% of equity per trade (0.08R)

Impulse trades

~112 extra per trader, also on missed signals

Half strength

10% missed, 20% early, 15% stops moved, 1 impulse per 8

Edge per signal

total R incl. impulse trades and costs ÷ 450

RNG

mulberry32, seed 20261015, same for all scenarios

This is a model, not a backtest of real prices. After the first leg, the random-after-entry odds are those of a fair game between the same price levels; trending doubles and snapping back halves the odds of a move continuing. The first leg is set so that every model gives 40% winners for the rules as written. Within one market model, every scenario trades exactly the same signals; across models, the paths differ. The habit rates, the −2.5R give-up level and the flat cost per trade are simplifying assumptions, and real slippage on blown stops would make moved stops worse. Every number comes from sim-discipline.js (plain Node.js: node sim-discipline.js, about two minutes). Change the assumptions and rerun it.