What a Year of Journaling in R Actually Shows

Ticker, date, dollars in, dollars out, a sentence about why you took it. That’s the journal almost everyone keeps the first time round, and it gets abandoned after a few months for a reason that nobody wants to say out loud. Reading it back teaches you nothing. Some trades won, some lost, and every sentence still sounds sensible.

The second attempt changes one column. Every result goes in as a multiple of what you risked, so a full stop-out is −1R, a trade that made twice the risk is +2R, and a year of that, with nothing skipped and nothing rounded in your favour, produces lessons that tend to arrive in roughly this order.

Most traders keep the first one because somebody made them. The habit usually starts when a mentor, a desk head or a trading group asks “and what did you actually do?” often enough that there is nowhere left to stand. It gets resented for about six weeks. Then the numbers start answering questions that had been argued in circles for a year, and the resentment goes.

Month one: the losses aren’t −1R

The plan says a loss is a planned loss, somewhere near −1R. Your first thirty rows will argue. Losses of −1.4R, −1.8R, one at −2.6R.

In dollars these look like ordinary bad trades, and that is exactly the trouble, because a $340 loss and a $210 loss sit in a dollar journal as the same species of event, with nothing in either row to tell you that one of them was the plan working and the other was the plan being abandoned halfway through. R separates them on sight. The −2.6R was a stop that got moved. The −1.8R was a mental stop you argued with and lost. The −1.4R was a gap. Gaps happen. The other two are yours.

So the first lesson lands inside a month, and it has nothing to do with losing. Losing is the job. The problem is losing more than you agreed to lose, and a dollar column will hide that from you for as long as you keep one.

Dollars record outcomes. R records outcomes against intent, and the space between those two is where every execution problem you have is sitting.

Month three: the winners are getting cut

Losses under control, the wins start telling on you. Average winner around +1.3R. The targets marked on those same charts averaged about +2.5R.

The gap is early exits. Not every time, but often enough to eat the edge, and the reasons are sitting right there in the notes once you read them together: took profit before earnings, looked heavy, didn’t want to give it back. Each one is defensible. Read as a batch, they describe somebody who picks a target off the chart and then overrides it with their stomach a few hours later.

Partial exits are the usual patch. Half the position comes off at a fixed 1.5R and the rest rides to the chart target or a trailing stop. It maximises nothing. What it does is give the nervous part of you a job small enough that the rest of the position lives through the wobble that would otherwise have closed it.

Month six: the setup breakdown

A hundred and twenty rows or so, which is enough to cut the results by setup. Say you run three: pullbacks in trends, breakouts from consolidation, reversals at levels.

Here is the breakdown that keeps turning up. Pullbacks carry a clear positive expectancy, breakouts land roughly break-even once costs are in, and the reversals are negative by a margin that stings.

And you’d have sworn reversals were your best work. They make the trades you tell people about, because a reversal that comes good is dramatic and a pullback that comes good is a Tuesday. Memory keeps the drama. The journal keeps the arithmetic, and dropping the reversals off the list tends to move your overall expectancy further than anybody would have guessed before they sat down and did the sum.

End of year: the number and its error bars

A little over two hundred trades logged, and an expectancy of, say, +0.3R per trade. Everybody wants that number, and hardly anybody asks how much of it is noise.

The statistics guides here go through confidence intervals properly, but the short of it is that given the spread of R values a normal log contains, the honest range around +0.3R after two hundred trades runs from roughly zero to a bit over half an R. A year of work. The data still can’t rule out that the edge is tiny. It can rule out negative, which counts for something, though the precision you’d quietly been assuming was never in the sample to begin with.

Sizing has to answer to that. If you can’t say whether your edge is 0.1R or 0.5R, then sizing as though it’s 0.5R is a bet on the friendly end of your own error bars. The expectancy and position sizing guide has the arithmetic. The journal is what makes you care about it.

The three columns that matter

Strip the thing back to three columns and keep these:

1. Result in R, so every trade is comparable and anything worse than −1R stands out on sight.
2. Setup name, one word, so that a hundred trades later you can see which setups are paying for the others.
3. Reason for the exit, a few words, because when the average winner keeps coming in under the average target you’ll want to know what you were thinking at the time.

The rest of what you could log is interesting and changes nothing. Those three change how you trade, and the stock profit calculator will turn a closed trade into R for you as long as you feed it the original stop, which removes the last excuse anyone has for not doing it.