forex-scalping.com

I finally had 500 tagged EURUSD and GBPUSD scalps since February with clean R values, so I plotted the distribution. The chart is the most useful single picture of my trading I've ever made, and I'd encourage anyone with a decent sample to do the same.

The shape, described in words because it's simple enough not to need the image.

A tall cluster of losers sitting at exactly minus 1R, about 31% of trades. Those are stops hit cleanly.

A smaller group between minus 1.1R and minus 1.5R, about 4% of trades. Slippage, mostly around news.

A group around zero, between minus 0.2R and plus 0.2R, about 18%. Scratches and break-even exits.

Winners spread from 0.5R to about 2R, with a peak around 1.2R, about 40% of trades.

A thin tail out to 4R and beyond, about 7%. Those trades produced roughly a third of my total profit.

What it told me.

The scratches are a bigger group than I'd realised. Nearly one trade in five goes nowhere, and each one costs spread and commission.

The right tail matters enormously. If I cut my winners at 2R, I'd lose a third of my profit.

The slippage losses are small in number but painful. Most come from trades within 15 minutes of a release.

Expectancy for the whole sample was about plus 0.19R per trade. Taking out the 4R tail, it drops to plus 0.07R. Without the slippage losses, it rises to plus 0.24R.

I redrew your buckets as a histogram from the percentages you gave, attached as an illustration, because seeing the gap between the minus 1R spike and the scratch pile makes the point better than any table.
The scratch group is the one I'd attack first. If each scratch costs around 0.1R in spread and commission, 18% of trades at 0.1R is about 0.018R per trade across the sample, roughly a tenth of your expectancy, spent on trades that went nowhere.

Quick sanity check on the arithmetic though. 40% of trades around 1.2R plus 7% out near 4R, minus 31% at 1R and 4% at roughly 1.3R, comes to about plus 0.4R per trade, not 0.19R.

So I'm guessing either the winner cluster averages well below the 1.2R peak, or the tail sits nearer 2.5R to 3R for most of those trades. Either one would change which fix matters most.

The slippage numbers do line up for me. Twenty trades at about minus 1.3R removed is roughly 26R over 500, which is your jump from 0.19R to 0.24R.

Are the R values after commission, or is cost tracked separately?

LondonScalper, one test I ran on my own log after seeing your tail numbers is removing the best trades one group at a time. It shows how fragile an edge is far better than the overall expectancy figure does.

On 400 tagged scalps my expectancy was plus 0.15R, about 60R in total. Taking out the best 5 trades dropped it to plus 0.10R, the best 10 to plus 0.06R, and the best 20 to roughly flat at plus 0.01R. The chart below is an illustration of that check on my sample.
Twenty trades out of 400 is 5% of the sample carrying the whole result. That is not a reason to panic, but it does mean those trades deserve their own tag sheet.

When I tagged mine, 14 of the 20 came from the first 90 minutes of London on days with a tight Asia range. That told me to protect that window and stop trading the afternoon, not to change my exits.

Did your 4R tail trades cluster on particular days or sessions, or were they spread evenly across the 500? If they cluster, the session filter may matter more than the exit rule.