Defining Win Rate for Trades, Positions, and Trading Periods
Summary
The discussion considers how to calculate a trader's win rate when the trader may finish each day flat, but may enter and exit positions through several executions. It distinguishes a period-based measure, such as profitable days, from a trade-based measure. The replies suggest that a trade is not necessarily identical to one order or fill: multiple buys and sells may represent one position, separate entries, or a strategy that adds to or reduces exposure. Trade grouping therefore depends on the trading process and on a consistent accounting convention.
The examples also show how frequent closing and reopening can inflate the reported proportion of winning trades without improving the economic result of the underlying position. Win rate should be interpreted alongside average win size, average loss or risk, and how the strategy defines a completed trade. The illustrative share sequence appears to contain a quantity inconsistency, and the answers do not provide a universal rule for allocating profits across partial executions.
Key ideas
- Win rate can be measured over trading periods or over defined trades.
- Order executions and completed trades are not necessarily the same unit.
- Position additions, partial fills, and exits affect how executions should be grouped.
- Closing and reopening can raise win rate while obscuring the result of a continuous position.
- Win rate should be assessed with win size and risk measures.
Tags
Full text
# Calculating win-ratio # Calculating win-ratio I want to quantify the performance of a trader and as I see in the books, there's a performance measurement called win ratio for this. In a book, it is defined as: ``` Win ratio explains what portion of the trades, trading days or trading months ended profitably. WinRatio = # Trading Periods|Gain>0 / # Total Trading Periods ``` Let's think of a day trader. She closes the day as flat. So, in order to define win ratio for her, I must calculate trade-by-trade performance, right? Because she is always flat at the end of days(my period is 1 day). For her, is it reasonable to define win ratio like the following? ``` WinRatio = # Trades|Gain>0 / # Total Trades ``` If so, then I want to clarify how should I count the winning and losing trades. Let's say she made the following operations: - Operation 1 - bought 10 shares at €3 - Operation 2 - bought 10 shares at €4 - Operation 3 - sold 3 of 20 shares at €2 - Operation 4 - sold 5 of 17 shares at €1 - Operation 5 - sold 2 of 12 shares at €4 - Operation 6 - sold 12 shares at €3.6 Should I consider both buy and sell as separate trades? I mean, is it correct that there are 6 trades here? or only the sell operations should be considered as trades? In my opinion, Operation 1 and Operation 2 are not trades, is it correct? If so, how should I tag these operations as "Winning trade" or "Losing Trade"? ## Answer by user42108 (score 1) https://quant.stackexchange.com/a/67962 Win rate is normally 'per trade' rather than 'per period'. "is it correct that there are 6 trades here?" - your fictional trader seems to have bought 20 shares and sold 22. I assume that's a typo and if so, I'd say you have either one trade (the net of all P&S) or two trades (as there are two buy-to-open executions). Is the strategy pyramiding? averaging down? Getting partial fills? That should affect how you think about the number of trades. ## Answer by brownie74 (score 0) https://quant.stackexchange.com/a/68410 Win rate can be tricked. I tweaked my strategy last week to lock in profits, and it did that by closing out and re-entering the trade immediately. So, through this 6 bar trend, i had 3/3 winning trades and no losses. I only made 1 other trade (loss) that day. So my win rate was 3 / 4 = 0.75, but really my win rate was 0.5 because the 3/3 was really one trade but i kept getting in and out. You have to consider your average win amount, versus your average risk (stoploss size expressed in money), alongside win rate. If you are risking a lot to make very little, and you have a winrate < 0.5 ... something is wrong. Because you are taking a big risk (50/50) with a lot of cash, to make only a little return.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.