Scalp vs Runner Classification After Entry — Trade Management Question
jankela:
I do not want to disclose the indicator details. What I am looking for is advice from people...
I do not want to disclose the indicator details. What I am looking for is advice from people...
🙄
I hit the same problem on gold, M1 and M3 move so fast that most signals either run or die in the first couple of minutes, so I stopped trying to classify them in real time and just let the first few minutes decide.
What worked for me was a simple proof window. I give a trade three minutes after entry. If it has not reached +1R by then, or if it touches +1R and comes straight back to entry, I close it or move the stop to breakeven and take what is there. If it reaches +1R cleanly and holds above it for a minute or so, I let the rest run with the stop under the last M1 swing.
I tested it the boring way, logged MFE and MAE for the first five minutes on every signal, then checked how often a trade that failed the +1R test in three minutes ever became a runner. On my data it almost never did, so the rule cost me very little and saved a lot of give-backs. Your numbers might be different, that is the part worth measuring on your own signals before you hard code it.
I’d start the clock at the fill and define what counts as a runner from the checkpoint onwards. For example, does price move a set distance further in your favour before hitting the original stop, within a fixed time? For trades still open at the checkpoint, group them by which of your positive and negative levels was touched first, including a group for neither. Then compare how often each group meets your runner definition.
Choose the levels and times on earlier data and freeze them before testing a later period. Run the whole EA both ways on that later period, with the same entry logic and realistic costs, once with your current exits and once with the new rule. Keep trades that closed before the checkpoint in the overall results. An early exit can free up a slot for another trade, so compare net P/L and drawdown across the full runs.
Hello everyone,
I am working on automating a short-term XAUUSD system on M3/M1, and the problem I am dealing with is not really “how to find a signal”, but rather how to classify and manage the trade after the signal appears.
The signal itself often catches a real move. In tick/backtest analysis, many signals move quickly into profit, often around +1 to +2 dollars, and some of them continue much further and become real runners. The problem is that these cases are hard to distinguish in real time.
There are several types of trades:
some signals become real runners and need room to breathe;
some signals only give a quick small profit and then reverse;
some signals enter a choppy zone and should be closed with a small profit or a small loss;
if I treat all signals the same way, I either take profit too early on runners or give back too much profit in chop.
I am not looking for a ready-made strategy or a new entry signal. The signal already exists, and I do not want to disclose the indicator details. What I am looking for is advice from people who have dealt with a similar trade management problem.
How would you objectively classify a trade during the first few minutes after entry?
For example, I am considering metrics such as:
whether price hits +1 before hitting -1;
I am not asking anyone to build the system for me. I am mainly interested in how you would mathematically or statistically define the moment when a trade should switch from “scalp mode” to “runner mode”.
how quickly the first profit appears;
MAE before the first +1 or +2 move;
whether price returns to entry after reaching +1;
whether there is clean follow-through on M1/M3;
whether an opposite signal appears shortly after entry;
when to allow runner mode and when to simply lock a small profit.
The goal is not to capture the entire maximum move. If a day has a large total MFE potential, I would be happy to capture a reasonable part of it, for example 30–40%, but in a stable way. The main problem is defining a rule that separates “quick scalp/chop” from “runner” without overfitting.
Would you approach this through:
a time-based proof window;
MFE/MAE ratio in the first few minutes;
break-even / step-lock logic;
volatility-based adaptive trailing;
structural SL/trailing based on M1/M3;
or another method?
The attached images show the typical issue: the same type of signal often produces a move, but not every trade deserves the same exit logic. Some should be locked quickly, some should be allowed to run, and some should be aborted early.
Any advice on testing methodology and real-time classification would be appreciated.
I am not asking anyone to build the system for me. I am mainly interested in how you would mathematically or statistically define the moment when a trade should switch from “scalp mode” to “runner mode”.