ForgeXBacktesting
Why your backtest beat your live results
Almost everyone trades worse live than in testing. Here are the seven reasons why, and which of them you can actually fix.
Your test said plus 40 percent. Six months live you are down. This is the most common experience in trading and it is not usually bad luck. It is a list of specific, findable differences between the test and the real thing.
1. You left out the costs
Spread, commission and slippage come out of every trade. On a strategy that averages a small win, they are the difference between profit and loss.
Work it out. If you make 8 points a trade and pay 2 points in spread, a quarter of your edge is gone before you start. If your test ignored that, your test was wrong by a quarter.
This is the easiest one to fix. Put real costs into the test and rerun it.
2. You skipped the bad ones
In a backtest you take every signal because you are working through a list. Live, you see a signal at an awkward moment and something in you says not this one.
The trades you skip are not random. People skip after a loss, when they are unsure, and when the setup is ugly. Those are often the ones that pay.
The way to catch this is to log every signal you saw, not just the ones you took, then compare.
3. You moved your stop
In the test the stop is a line and price either hits it or does not. Live, the stop is a decision you get to change while you are losing money.
Moving a stop turns a controlled loss into an uncontrolled one, and it usually works several times before the once that hurts.
4. Your size moved around
Testing usually assumes constant risk. Real traders size up after a run of wins and either freeze or double up after losses.
The result is that your biggest positions land on your worst trades. That alone can turn a profitable set of trades into a losing account.
5. You saw the future without meaning to
If you tested by scrolling a chart, you knew what happened next. You cannot switch that off by trying.
Bar by bar replay is the fix, because the future is physically not on the screen.
6. Not enough trades to begin with
If the test was forty trades, the live result is not a contradiction of it. Both are small samples and they were always going to disagree.
7. The market changed
Sometimes the edge was real and then it went away. Other people found it, or conditions moved.
This is the only one on the list you cannot fix, and it is also the rarest. Work through the other six first, because the honest answer is usually in there.
How to find out which one it is
Put your live trades and your tested trades side by side and compare four things. Average result in R. Win rate. Average holding time. Average size.
Whichever one differs most is your answer. Holding time is the most revealing, because it usually shows losers being held longer live than in the test, and that is a discipline problem rather than a strategy problem.