ORB Trading Metrics

There is a predictable arc. A trader decides to take measurement seriously, builds a spreadsheet with a great many columns, fills it in meticulously for two weeks, then sporadically, then not at all. Three months later the file is opened, found to contain a detailed fortnight followed by scattered fragments, and abandoned. The instinct afterwards is to build a better one. The problem was never the design of the fields.

The Real Constraint Is a Bad Session

Any tracking system will be maintained on a good day. The question is whether it gets filled in after a session that went badly, when the last thing anyone wants is to sit and document exactly how.

That is the moment the system is designed for, and it is the moment most designs fail. A thirty field form takes long enough that skipping it once feels reasonable, and the gap in the record then makes the next entry feel pointless. Systems die from a single skipped day far more often than from a decision to stop.

Partial Records Mislead in a Specific Direction

An incomplete record is not simply less useful than a complete one. It is biased, and biased predictably, because the sessions that go unrecorded are not a random sample. They are disproportionately the difficult ones.

So the surviving record over-represents sessions where things went smoothly and the trader had the energy to document them. Any conclusion drawn from it about adherence, about frequency, or about which conditions suit the approach will be flattering and wrong. A small record covering every session is worth considerably more than a rich record covering the comfortable ones.

Most Fields Are Never Read

The second failure is that elaborate systems collect data nobody consults. Fields get added because they might be interesting, not because a specific question needs answering, and the collection cost is paid daily while the value is theoretical.

A useful discipline is to require that every field be attached to a question you would actually act on. If the answer to which day of the week performs best would not change anything you do, the column is costing time and returning nothing. Most abandoned dashboards are three quarters columns of this type, and their weight is what killed the quarter that mattered.

What Survives

A system that lasts tends to be small enough to complete in a minute or two and simple enough to fill in on a bad day. In practice that means a handful of fields per session, not per trade, with per trade detail limited to what can be jotted quickly.

Something like the date, how many signals appeared, how many were taken, whether size matched the plan, whether the exits were planned or improvised, and one short line of description. That is enough to support most of the questions worth asking, and it is short enough that filling it in never becomes a decision.

The form matters less than the friction. Paper works. A note on a phone works. A spreadsheet works. What does not work is anything requiring a process to be started, a file to be found, or fields to be looked up.

Growing It Deliberately

None of this argues against ever adding detail. It argues for adding it in response to a question rather than in anticipation of one.

When a pattern appears that the current record cannot explain, that is the moment to add a field, and the field should be the narrowest one that answers the question. Once it has been answered, the field can usually be removed again. Treating the tracking system as something that expands and contracts around live questions keeps it small, and small is the only property that reliably predicts whether it will still exist in a year.

The measure of a tracking system is not how much it captures. It is how many consecutive sessions it has captured without a gap, and a modest system with an unbroken run is a genuinely useful record while an ambitious one with holes is mostly a reminder of an intention.

Almost everyone believes they follow their rules. The belief survives because deviations are small, each one had a reason at the time, and reasons are far more memorable than the deviations they justified. Measuring adherence is therefore not a matter of being honest with yourself in the abstract. It is a matter of recording something before memory has a chance to smooth it over.

A Rule You Cannot Score Is Not a Rule

The first obstacle is usually the rules themselves. A rule reading enter on a convincing break cannot be scored, because convincing was never defined and any entry can be described as having met it after the fact. The same applies to exit when the move looks exhausted or reduce size in poor conditions.

Making rules checkable means writing them so that a person who was not there could look at the chart and the record and say yes or no. That usually forces uncomfortable specificity, which is the point. Rules written vaguely are not being kind to your judgement, they are avoiding the moment when the judgement gets evaluated.

Score at the Time, Not at the Weekend

Adherence recorded later is adherence remembered, and memory is systematically generous about decisions that felt reasonable. The reconstruction is not deliberate. A trade entered late becomes a trade entered when the setup matured. A skipped signal becomes a signal that did not really qualify.

The remedy is to note it immediately, in whatever crude form survives the pressure of a live session. A single character next to each trade is enough. What matters is that the note was made while the decision was still fresh, before the outcome is known, because the outcome exerts a strong pull on how the decision gets remembered.

Recording Deviations, Not Just Counting Them

A count of deviations tells you how often. A brief note of what kind tells you what to fix. Entering before the trigger, entering after it, taking a signal that did not qualify, skipping one that did, sizing differently, moving a stop, exiting early, exiting late. Half a dozen categories cover most of what happens.

Patterns emerge from the categories rather than the total. Someone whose deviations are almost all early entries has a different problem from someone whose deviations are almost all skipped signals, and the second person may be closer to trouble despite doing less. Skipping is quieter, feels prudent, and hollows out a strategy without producing a single bad trade to point at.

The Distinction Adherence Buys You

The reason all of this is worth doing arrives during a losing stretch, which is when the important question gets asked: is the strategy wrong or was it not run?

Without an adherence record there is no way to answer. The trader looks at a poor month and can construct either story, and will generally construct the one that requires less discomfort. That is usually the one blaming the strategy, because changing a strategy is easier than changing a habit.

With a record the answer is available. A losing month with clean adherence is evidence about the strategy and should be treated as such. A losing month with a dozen recorded deviations is evidence about execution, and changing the strategy in response would be solving the wrong problem while destroying the only clean data you had.

Reviewing Without Flogging Yourself

A record of deviations is easy to turn into a source of guilt, which is counterproductive because the guilt eventually makes the record unpleasant to maintain and the record stops.

The more durable framing treats deviations as information about where the rules and the person do not fit together. A rule that is deviated from constantly may be a bad rule, or one that asks for something unrealistic given how you actually trade. Sometimes the right response to a persistent deviation is to change the rule to match what you keep doing, provided that behaviour holds up under examination. Sometimes it is to change the behaviour. The record does not decide which, but it is the only thing that puts the choice in front of you clearly.

Ask a trader how they are doing and the answer is a number about money. It is the natural answer and it is also the one with the worst signal to noise ratio available. Over a short window, results are dominated by which sessions happened to occur, not by what was done during them. A behaviour metric has the opposite property: it reports on the part you controlled, and it reports quickly.

Why Outcome Is Slow to Say Anything

A breakout approach produces a scattered distribution of results. Most trades are small either way and occasional ones are large. A handful of sessions can dominate a month, which means a month is not enough data to distinguish a good process from a lucky one.

Worse, the feedback arrives with the wrong labels attached. A trade taken carelessly that worked feels like confirmation. A trade taken properly that failed feels like a mistake to be corrected. Learning from outcomes over short windows therefore teaches the wrong lesson roughly as often as the right one, and it does so persuasively because money is involved.

Signal Count and Participation

The simplest behavioural measure is how many valid signals appeared and how many were taken. Both numbers are countable at the end of a session without any reference to whether the trades made money.

The gap between them is informative on its own. A trader taking a small fraction of the signals their own rules generated is running a different strategy from the one they think they are running, and whatever the results say, those results are not a test of the written rule. The reasons for skipping are worth a short note as well, because they cluster: hesitation after a loss, distraction, and a view about the session are all common and all different problems.

Delay Between Trigger and Action

Time from signal to order is a small measurement with a large amount of information in it. It is easy to capture, needing only a rough note of when the level broke and when the order went in.

Consistent delay usually indicates uncertainty about the rule rather than a technical problem. Delay that varies with recent results is more revealing still, because it means the previous trade is influencing the current one, which is the definition of an undisciplined sequence. This is the sort of pattern that is completely invisible in a profit and loss record and obvious after a fortnight of timestamps.

Size Consistency and How Trades End

Recording planned size against actual size takes seconds and catches one of the most damaging habits there is. Size that quietly increases after wins and shrinks after losses produces a distribution of results far worse than the strategy itself would, because the largest positions are concentrated in the periods most likely to be followed by a reversal.

Nobody sets out to do this and a great many people do it. It is not visible in an average, since the average size may be exactly as planned. It is only visible when size is recorded alongside the sequence.

Every closed position can be labelled with how it ended: the planned target, the planned stop, or something else. That third category is the one worth watching. It covers exits taken because the trade felt wrong, because a level looked close enough, or because attention was needed elsewhere.

A rising proportion of improvised exits indicates the plan is being overridden regularly, and it usually appears before the results deteriorate. Because the label is applied at the time of the exit rather than reconstructed later, it resists the retrospective tidying that memory performs on uncomfortable decisions.

What Makes These Worth the Effort

Behaviour metrics share three properties that outcome metrics lack. They generate data on every session, including the ones with no trades. They point at something specific enough to change. And they cannot be flattered by a favourable stretch of market conditions, because they never referred to the market in the first place.

None of this argues for ignoring results. Results are the eventual arbiter and they should be recorded. The argument is about sequence: behaviour first, because it is what you can act on now, and because a result you cannot attribute to anything is a number rather than a lesson.