Forex R-Multiple & Win/Loss Correlation Table | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

MT4 MTF Correlation table and Correlation oscillator indicators 155 replies

Multiple Timeframe & multiple currency of RSI alert 0 replies

Correlation Table - Importing OHLC Data for all pairs into a spreadsheet. 3 replies

Multiple Time Frames & Multiple Markets 3 replies

Correlation Table current looking 1 reply

Attachments: Forex R-Multiple & Win/Loss Correlation Table
Exit Attachments

Forex R-Multiple & Win/Loss Correlation Table

  • #1
  • First Post: Mar 19, 2024 9:01pm Mar 19, 2024 9:01pm
Figure I'd post on FF as well.

Ran through these Risk-Reward scenarios to show their win %. These numbers are self-evident; the building blocks to a quantitative strategy. Use as a reference.

R Value(Reward/Risk) - Expected win %

0.02 - 97.7%
0.05 - 95.0%
0.1 - 90.0% (10:1)
0.15 - 86.4%
0.2 - 83.3% (5:1)
0.25 - 80.0%
0.3 - 76.0%
0.35 - 73.3%
0.4 - 71.2%
0.5 - 66.6% (2:1)
0.6 - 61.5%
0.7 - 57.7%
0.8 - 54.0%
0.9 - 52.0%
1.0 - 50.0% (1:1 Risk:Reward)
1.1 - 47.0%
1.2 - 45.0%
1.3 - 43.0%
1.4 - 41.3%
1.5 - 40.0%
1.7 - 36.0%
2.0 - 33.3% (1:2)
2.5 - 28.3%
3.0 - 25.0% (1:3)
3.5 - 22.7%
4.0 - 20.0%
5.0 - 16.6%
6.0 - 14.3%
7.0 - 12.5%
8.0 - 11.4%
9.0 - 10.4%
10. - 9.5% (1:10)
12. - 7.9%
15. - 6.5%
30. - 4.1%
37. - 3.8%
40. - 3.5%
50. - 2.9%

All scenarios taken from EU 2018-2024, min. 10,000 trades.
ForexOrderFlow.com Return This Year: 15.7%
  • #2
  • Jun 7, 2024 3:46pm Jun 7, 2024 3:46pm
There is no need to measure this. If you assume the market is a random walk, you can simply calculate using:

Inserted Code
winrate = 100% * (1 - RR/(RR+1))

For example:

Inserted Code
RR = 5.0
winrate = 100% * (1 - 5.0/6.0) = 16.7%

Hope this helps. Good luck
  • #3
  • Jun 7, 2024 4:14pm Jun 7, 2024 4:14pm
Quoting yoriz
There is no need to measure this. If you assume the market is a random walk, you can simply calculate using: winrate = 100% * (1 - RR/(RR+1)) For example: RR = 5.0 winrate = 100% * (1 - 5.0/6.0) = 16.7% Hope this helps. Good luck
Agreed it is quantifiable. These are real results since I have a dedicated PC on the MQL5 cloud now.
ForexOrderFlow.com Return This Year: 15.7%
  • #4
  • Jun 7, 2024 4:21pm Jun 7, 2024 4:21pm
Quoting Ryan420
{quote} Agreed it is quantifiable. These are real results since I have a dedicated PC on the MQL5 cloud now.
Always good to double check your assumptions. Nice that your measurements confirmed this. Only at the extreme ends of your table, the math doesn't add up, but I assume these are just noise measurements with very little samples.
  • #5
  • Jun 7, 2024 4:29pm Jun 7, 2024 4:29pm
Quoting yoriz
{quote} Always good to double check your assumptions. Nice that your measurements confirmed this. Only at the extreme ends of your table, the math doesn't add up, but I assume these are just noise measurements with very little samples.
Oh, I'd be interested in bridging the gap as to why. Since I have this in Excel, I made it in to a graph to plot the slope
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot_20240607_142803_Sheets.jpg Size: 62 KB
ForexOrderFlow.com Return This Year: 15.7%
  • #6
  • Jun 9, 2024 10:20am Jun 9, 2024 10:20am
Quoting Ryan420
Oh, I'd be interested in bridging the gap as to why.
At the extremes you get measurement errors because either the TP or SL is tiny. This means your backtest simulation is extra sensitive to slippage, spread, commission because the position is closed in only a few ticks after opening.

Quoting Ryan420
Since I have this in Excel, I made it in to a graph to plot the slope
You could perhaps add a line from the formula I posted for reference to double check you measurements follow that line?
  • #7
  • Jun 19, 2024 7:42am Jun 19, 2024 7:42am
Quoting yoriz
{quote} At the extremes you get measurement errors because either the TP or SL is tiny. This means your backtest simulation is extra sensitive to slippage, spread, commission because the position is closed in only a few ticks after opening. {quote} You could perhaps add a line from the formula I posted for reference to double check you measurements follow that line?
Here's the chart plots compared. Red dots are my results, blue dots are the formula's 1-(R/(1+R))*100

Most noticeable is the loss in accuracy when going very high R-multiples.
Attached Image (click to enlarge)
Click to Enlarge Name: Untitled.png Size: 17 KB
ForexOrderFlow.com Return This Year: 15.7%
  • #8
  • Edited 5:10am Jun 20, 2024 1:45am | Edited 5:10am
Good to see the dots nicely match!

I am curious how you will now apply this knowledge about the win rate to your strategy. Vary the lot sizes based on this? Vary the SL/TP levels?
  • #9
  • Jun 20, 2024 3:55am Jun 20, 2024 3:55am
Quoting yoriz
Good to see the dots nicely match! I am curious how you will now apply this knowledge about the win rate to your strategy. Vary the lot sizes bases on this? Vary the SL/TP levels?
When running all the R-multiples, I was able to get snapshots of their equity charts. Too low of an R would create a very slow equity slope up, with large drops every now and then when a loss comes. Too high of an R, you'll drift down with losing trades and spike up when the eventual winner hits. Both scenarios were unsustainable. The sweet spot is somewhere between them.
I also concluded that R above 1 would always have a longer losing streak than winning streak.. inversely below 1 would have longer win streaks than losing streaks.

I settled on using 0.8R (54% win rate.. or 55.6% in the calculation) as a 40 pip TP & 50 pip Invalidation as my base. It's enough below 1 to guarantee win streaks are longer than losers, and high enough to be able to double my bet if price moves against me.

Since price does not matter, I can visually simulate trading like so: If EurUsd makes a low at 1.0700, I'd like to enter when there is room for two of these trades to lose back to back. Being 50 pips is invalidation, I'll start looking for shorts about 100 pips away from the low, so 1.0800 or just beyond its highs. I still prefer to enter where those shorts would put their stops. I'll enter with a 40 pip TP.. if price goes against me, then I'll look to enter double the amount at a level at least 50 pips away. The 2nd trade will have a TP of 25 pips with a new invalidation level now 100 pips away. An R-multiple of 0.25, or 80% win rate. If it gets to 25 pips, I exit both. I go -15 pips on the first frade and +25 on the 2nd trade which is double in size (+50 really) for a net of +35 pips. Slightly lower than the +40 original entry. If price hits that loser 20% of the time and goes 100 pips without a retracement, I'll double again and enter with a 50 pip TP to get all 3 positions to net breakeven. I've yet to enter that 3rd time. I'll know more when I get there.

I started tracking these results in Demo, and started following them Live. Here's the thread for this: https://www.forexfactory.com/thread/post/14886926
ForexOrderFlow.com Return This Year: 15.7%
  • #10
  • Jun 20, 2024 5:20am Jun 20, 2024 5:20am
Quoting Ryan420
I started tracking these results in Demo, and started following them Live. Here's the thread for this: https://www.forexfactory.com/thread/post/14886926
Looking forward to see your results! Are you trading manually or did you automate this? If so, have you run a backtest over a few years to see what drawdowns to expect?
  • #11
  • Jun 20, 2024 5:33am Jun 20, 2024 5:33am
Quoting yoriz
{quote} Looking forward to see your results! Are you trading manually or did you automate this? If so, have you run a backtest over a few years to see what drawdowns to expect?
It's a hybrid.. I'm trading very similar to the results. The thing is with the results is that they were all "random" entries (Bollinger Band on 1 min TF). I noticed some really bad entries lead to horrible results. These were shorts taken at relative lows and buys taken at highs. Since price tends to retrace for more liquidity rather than explore new extremes to find it, I decided to trade retracements, support and resistance.

The model will fail eventually. The key will be to at least double an acct and start a new one before it happens. The probability of it failing will be the expected loss % from the 3 entries times each other. So 46% x 20% x 20% = 1.8% If I can make a 4th entry, it'll drop the risk of ruin to 0.4%. I'll have to calculate the inevitable drawdown though.
ForexOrderFlow.com Return This Year: 15.7%
  • #12
  • Jun 20, 2024 6:54am Jun 20, 2024 6:54am
Quoting Ryan420
The probability of it failing will be the expected loss % from the 3 entries times each other. So 46% x 20% x 20% = 1.8%
If your intend is to double your account before that time, you could look at the average amount of iterations before you ruin your account. At 1.8% that would be:

Inserted Code
50% = (100 - 1.8%)^n
n = log(0.5)/log(0.982) = 38 trades

So in 38 trades (on average) you must have doubled your account. That means the profit of each iteration (aka group of trades) must be at least:

Inserted Code
(1 + return)^38 = 2.0
return = pow(2.0, 1/38) - 1 = 1.84%

You could use that to size your trades.

I am usually not so into these "everything on Red" casino-style strategies, but I find it interesting to see how this performs. I'll try to make time to code this into an EA and see what drawdowns backtesting gives.
  • #13
  • Jun 20, 2024 9:35am Jun 20, 2024 9:35am
Quoting yoriz
{quote} That means the profit of each iteration (aka group of trades) must be at least: (1 + return)^38 = 2.0 return = pow(2.0, 1/38) - 1 = 1.84% You could use that to size your trades.
Exactly what I'm looking for. Currently that demo is at +0.9% average per trade incl the losers. You're saying I need to net at least +1.84% per group then? So each single winner then must be +1.84% = 40 pips TP and I'll have to add a little sugar more than double on to the 2nd trade entry.

I cut my teeth in a casino. It is totally a roulette wheel without the Zeros.
ForexOrderFlow.com Return This Year: 15.7%
  • #14
  • Jun 20, 2024 10:03am Jun 20, 2024 10:03am
Quoting Ryan420
Currently that demo is at +0.9% average per trade incl the losers. You're saying I need to net at least +1.84% per group then?
You named it "risk of ruin", so I based the calculation on the assumption that you will be ruined if you meet the unlikely series of events of that 1.8%. That implies you take lot sizes that are huge and represent using up the maximum margin / deposit load in case you need to scale-in all three trades of the group. At that lot size, the average gain per trade (or group of trades if you need to scale in) should be at least +1.84%.

In reality you might want to tune down the volume a bit and make sure you can survive the Black Swan, although that probably will still incur a large drawdown. The best way to find the best sizing would be run a few backtests and see what typical drawdowns you can expect at certain lot sizes. Assuming mathematicians are right, we will not be able to find a lot size that (on average) doubles your account before you meet the black swan. In other words, we might take a gamble and get lucky, but it is not a sustainable strategy.

Quoting Ryan420
I cut my teeth in a casino. It is totally a roulette wheel without the Zeros.
No, it is a roulette wheel *with* the zeros. The zeros are the commission, spread and slippage that you pay for each trade, and the margin requirements that prevent you from going all-in.
  • #15
  • Jun 20, 2024 11:50am Jun 20, 2024 11:50am
No, not sustainable mathematically. I'm hoping my experience with order flow trading and market microstructure can help weed out the trade(s) that would else break me. As long as the wind is at my back, and I can trade better than random chance, the numbers should skew towards profitability. It'll be interesting to see.

I've researched many complex roulette betting systems in relation to forex. One I landed on that has a chance is called "Fibonacci Dozens". No relation to Fibs in trading.

I'll shoot to aim for 2% per trade coming up and see how it goes. I do have room in drawdown as I have never went below the starting acct balance.. yet.
ForexOrderFlow.com Return This Year: 15.7%
  • #16
  • Jun 20, 2024 12:19pm Jun 20, 2024 12:19pm
So if The acct is $6600, +2% would be $132. To do that with 2 positions and net +35 pips, it's 0.38 lot. The initial trade of +40 pips will have to be +$152 or +2.3%. I've round the lot up to .4 and will grow accordingly.
ForexOrderFlow.com Return This Year: 15.7%
  • #17
  • Edited Jun 21, 2024 2:00am Jun 20, 2024 5:19pm | Edited Jun 21, 2024 2:00am
I wrote an MT5 expert advisor to try out your idea. You mentioned you trade S&R levels, so I used a Williams R% at level 0 and -100 to trade the reversal. Your manual entries are probably better, but just for the experiment the Williams R% might work.

I optimized the number of lookback bars of William R% indicator over a four year period (2019-2022) on maximum Martin Ratio:
Attached Image (click to enlarge)
Click to Enlarge Name: screenshot.png Size: 10 KB

I ignored the few outliers, and selected the center of the highest plateau (encircled in red). That gives a lookback period of 175 H1 bars. I then traded this setting from 2023 forward:

Attached Image (click to enlarge)
Click to Enlarge Name: screenshot.png Size: 12 KB

I must say I am not disappointed. No Black Swans here!

However, the deposit load is already at maximum (with 1:500 leverage) so this is as good as it gets. With a return of about 50%/year you might need to rethink your idea to regularly double your account to avoid risk of ruin.
  • #18
  • Edited 7:17am Jun 21, 2024 5:55am | Edited 7:17am
I don't know if you are a MetaTrader 5 user, but here is the expert advisor I wrote should you want to play with it:
Attached File(s)
File Type: ex5 yzRyan420.ex5   38 KB | 233 downloads
  • #19
  • Jun 21, 2024 11:22am Jun 21, 2024 11:22am
Quoting yoriz
I don't know if you are a MetaTrader 5 user, but here is the expert advisor I wrote should you want to play with it: {file}
Yep, I'm all about mt5 & MQL5. Thank you so much! I will be on it all weekend.. what else do we do when markets are closed.

In the backtests above, what if we increased risk so that an acct would fail once or twice, but increase the ability for larger returns?
I'm not meaning to mimic or take a shortcut, but I follow Nick Shawn on YouTube. He has a very interesting and similar view on this process of risk management in which he will 5x an acct but blow one every so often.
ForexOrderFlow.com Return This Year: 15.7%
  • #20
  • Jun 21, 2024 11:51am Jun 21, 2024 11:51am
Quoting Ryan420
Yep, I'm all about mt5 & MQL5. Thank you so much!
Glad that it is useful to you! Entries will probably not be what you would have done manually, but good enough to play with the TP/SL/Volume settings to see what that does.

Quoting Ryan420
In the backtests above, what if we increased risk so that an acct would fail once or twice, but increase the ability for larger returns?
The backtest already eats up a lot of margin. If you increase the lot size any further, I don't think it can handle opening the 3rd trade of a group. You can see already 5 occassions in the above 18 month backtest where it reached level 3, so you would have to be extremely lucky to double your account before hitting the 3rd level.

However, don't let me influence your thinking too much and keep an open mind. Happy experimenting!
2 traders viewing now