## What is the winning rate of a trading strategy?

Hi community, I have a question about "Wining Rate".
I read a lot from different articles and find it confused with its definition.
• some ppl mentioned it is the percentage of trading day that end up with a gain
• some other ppl calculated it as the percentage of winning trades from a round order perspective

What is the correct definition and formula ?

Actually, both definitions are in use in the financial market.

Let's consider this example. Suppose we open a buy order of USDJPY at price 110 on day 0 with take profit level 120 and no stop loss input. Also, suppose the market closing price for the next following days are:

Day Closing Price Daily PnL
0 110 0
1 108 -2
2 105 -3
3 104 -1
4 100 -4
5 130 30
Total PnL 20

In this example, this trade has 4 consecutive loss days, but finally ends up with a large gain.

• Using trade based definition, as we only have 1 trade which is a profitable trade, so our win rate will be 100%.
• However, using the definition of "the percentage of trading day that end up with a gain", the win rate will be calculated as 1/5 = 20%.

Both answers are correct, but just present in different prespectives.

It's really a good question!

As you already mentioned, there are 2 different definitions in the market. Actually, many investors mix up the 2 concepts, particularly for retails. Intuitively speaking, a high win rate is preferable. However, when comparing the value across different investment funds/ strategies, we need to know whether their calculation methodology is consistent or not.

In fact, as seen from David's example, many poor traders/fund managers usually take the "trade base" approach. For profitable trades, they can close the orders to increase the number of profitable trades. However for losing trades, they will just leave it open and wait for market reverse. As long as the trader doesn't close the losing trade, those losing trade won't be counted toward the win rate calculation. Therefore, it is possible that a trader/fund manager presents a high value of win rate (eg. 90%), but the portfolio doesn't make profit.

Another thing is related to trading system design. For a trade based system (eg. MT4/MT5, etc) where orders are opened/closed as a round order manner, calculating win rate using a trade based approach is possible. However, for some position based trading system (eg. Interactive broker, Binance, etc), round order doesn't exists (eg. buying 10 shares can be unwound by selling 5 shares at 2 different time). Therefore, the trade based calculation is not generally applicable.

As a personal suggestion, win rate using the "profitable day" concept is more reliable and widely applicable to all kind of trading systems.

Original Posted by - David:

Actually, both definitions are in use in the financial market.

Let's consider this example. Suppose we open a buy order of USDJPY at price 110 on day 0 with take profit level 120 and no stop loss input. Also, suppose the market closing price for the next following days are:

Day Closing Price Daily PnL
0 110 0
1 108 -2
2 105 -3
3 104 -1
4 100 -4
5 130 30
Total PnL 20

In this example, this trade has 4 consecutive loss days, but finally ends up with a large gain.

• Using trade based definition, as we only have 1 trade which is a profitable trade, so our win rate will be 100%.
• However, using the definition of "the percentage of trading day that end up with a gain", the win rate will be calculated as 1/5 = 20%.

Both answers are correct, but just present in different prespectives.

I would to clarify a bit for the calculation using the "winning day" concept.
For below time series of daily PNL,

+1, 0, -1, +3, +2, -4, 0, +1, +1, +2

What will be the correct win rate?
• Count only positive number: 6/10 = 60%, or
• Count non-negative number: 8/10 = 80%

Original Posted by - Ka Ka: I would to clarify a bit for the calculation using the "winning day" concept.
For below time series of daily PNL,

+1, 0, -1, +3, +2, -4, 0, +1, +1, +2

What will be the correct win rate?
• Count only positive number: 6/10 = 60%, or
• Count non-negative number: 8/10 = 80%

Simply speaking, win rate counts for profit only (i.e. positive number in the time series). So, 60% is the correct answer.

But when you look into it further, you may find an interesting scenario.

What if a strategy inactive for long time?

The win rate will be decreasing!!!

So there is another modified calculation to ignore those inactive or zero PNL days. Then, the win rate will be 6/8 = 75%.