forex-scalping.com

How often is a profitable system still down after 20 trades? About one time in four

Two systems with positive expectancy. One wins 45% of the time at 1.5R, an expectancy of +0.125R. The other wins 40% at 2R, +0.2R. Simulated, 20,000 runs each, fixed risk and no costs, here's the chance the total is below zero after n trades.

After 20 trades: 25% for the first and 25% for the second.
After 50: 20% and 16%.
After 100: 14% and 9%.
After 200: about 6.5% for the first.

So a trader judging a working system after its first 20 trades has a one in four chance of seeing red. For a scalper that can be two days. For someone taking five trades a week it's a month.

It works in the other direction too. A system that loses, winning 45% at 1R with an expectancy of -0.1R, is in profit after 20 trades 25% of the time, 20% after 50 and 13% after 100.

Short samples can't separate a good system from a bad one very well. The first month of a new setup, positive or negative, says less than it feels like it does. I'm curious how others decide when a sample is long enough to act on, and whether you use a number of trades or something else.

One in four is a sobering number. It suggests judging a system by its first 20 trades is close to useless, and even 100 trades leaves a meaningful chance of a red result. For scalpers who take many trades, that sample is reached quickly, but swing traders may need months. Forward testing on a demo or tiny live size until a few hundred trades have accumulated is the patient way to avoid dropping a good system too early.

The sample size needed to tell the systems apart grows fast because the edge is small compared with the noise. For the 45% at 1.5R system, the standard deviation of one trade's result is about 1.25R and the expectancy is 0.125R, so the standard error after n trades is 1.25 over the square root of n. To get the standard error down to half the expectancy, about 0.06R, you need around 400 trades. That's roughly two years for someone who takes 4 trades a week, and a month for a scalper doing 20 a day. Fewer trades than that leaves a wide band around any measured result.

Calendar time depends on style. 20 trades is about a day for an active scalper, a month for a swing trader taking 5 a week. If a swing trader tests a new system for one month and stops after 20 trades in the red, there's a one in four chance that was luck, even if the edge is real. Scalpers can get the sample faster, but then they hit the other issue, that market conditions change, so 300 trades done over two weeks of low volatility may not predict the next two weeks. More trades isn't the same as more independent information.

One answer is to judge the first 50 trades on process and not on profit. Did you follow the entry rules, size correctly, and place the stop where the plan said? Count the trades where all three were true, and that gives a pass rate independent of luck. Profit and loss over those 50 trades can still be a long way from the long run average, with a 20% chance of red for the first example. After 100 trades with a clean process, the negative chance falls to about 14%, which is a better base for a decision, though still not a certainty.

The losing side deserves a look too. A weak system at 45% and 1R is ahead after 20 trades 25% of the time, so some early success is no proof either. A trader who raises size after a good first month on a system with no edge is making the opposite error. The fix is the same: plan the number of trades before any decision about size, write it down, and don't revisit it halfway through. Setting a size increase to need 100 trades with a positive expectancy and a stable drawdown guards against both mistakes.

Some brokers ban trades closed in under a certain number of seconds. Others accept them but reserve the right to cancel profits from "abusive" patterns. A few simply say scalping is welcome and mean it.

Typical language includes minimum holding periods of 30 to 60 seconds, limits on the number of orders per day, or clauses about "latency arbitrage" or "exploiting pricing errors." The wording is often vague enough to let the broker decide case by case.

For scalpers, the risk is real. A profitable account can be flagged, profits clawed back, or the account closed. The terms are usually buried in the client agreement, not on the marketing page.

Before depositing a meaningful amount, reading the clauses on trading conduct is wise. So is asking support directly and getting an answer in writing: are trades under a minute accepted? Is there a daily order limit? What happens if the account is flagged?

Brokers that welcome scalpers tend to state it clearly and publish execution statistics. Those that dodge the question are telling you something.

Some brokers apply the rule only to profitable trades. A trade closed at a loss after twenty seconds is accepted, but a winner closed at twenty seconds is voided. That asymmetry is hard to detect without looking carefully at the language.

Others measure the holding time in an unusual way: from the moment of order submission instead of the moment of fill, or including the time spent in a pending state. In that case, a trade that fills after a delay and closes quickly may violate the rule without the client realizing.

If the strategy depends on short holds, it is worth finding out exactly how the clause is enforced. Ask the support team for an example, and keep the reply.

Brokers that explicitly welcome scalping usually mention it in the account description and give examples of acceptable activity. The reassurance matters more than any spread advertisement, since a profitable account can't be withdrawn if the profits are cancelled.

What did the broker's written answer say about short-duration trades, and was it consistent with the terms?