Skip to content
Montex Alpha Rail System
$497
← Back to TEX — Expectancy

Operator brief · 396

Monte Carlo says what should happen. Expectancy says what is.

The key idea

Different tenses

One is conditional. The other is present.

The benchmark is a conditional statement: if the system continues to behave as its recorded evidence suggests, outcomes will distribute like this. Expectancy is a present-tense measurement: this is what the method is producing now, net of friction, in R. Neither can do the other's job. The benchmark cannot detect that something changed last month, because it was built before last month; expectancy cannot say whether its own reading is normal, because it has no distribution to sit inside. Keeping the tenses straight also prevents a subtle misuse: treating the benchmark as a forecast of the coming period rather than as a description of how a population of periods distributes.

The pairing

Each supplies exactly what the other lacks.

Together they answer a question neither can answer alone: is what is happening consistent with what should be happening? Expectancy supplies the live value; the benchmark supplies the range that value should fall within. This is the entire mechanism behind live-versus-simulation comparison, and it is why the first two syllables of the name are adjacent — they are one instrument with two halves rather than two features that happen to appear in sequence.

FigureWhat each half can see
MON — the benchmarkconditional, built in advance· The range outcomes should fallin· How deep adverse paths run· Whether a drawdown is normal· Blind to change since it wasbuiltTEX — expectancypresent tense, measured weekly· What the method produces now· Net of real friction, in R· Conditional by branch andregime· Blind to whether that value isnormalThe pairwhat only both can answer· Is live consistent withexpected?· Is this drawdown inside theenvelope?· Has the machine changed sincethe ruler was cut?· Is there excess, and is itattributable?

The pairing exists because the blind spots are complementary rather than overlapping.

Reading a disagreement

Live below the model has two very different causes.

When expectancy sits materially under what the benchmark predicted, exactly two explanations exist and they demand opposite responses. Either this is an unlucky ordering inside the expected distribution — in which case the correct action is to change nothing — or the system has genuinely changed since the benchmark was built, in which case the benchmark is stale and the method needs examining. Percentile position alone cannot separate them, which is why structural drift readings are consulted alongside: drift says whether the machine moved.

The other disagreement

Live above the model is the same problem with better feelings.

Outperformance decomposes identically: either a favourable ordering within the expected distribution, or the system has genuinely improved and the benchmark is stale in the other direction. Traders investigate the first case exhaustively and the second almost never, which is a consistent asymmetry and an expensive one — an out-of-date benchmark that is too pessimistic quietly under-authorises a system that has actually got better, and nobody files a complaint about a ruler that flatters them. Auditing outperformance with the same rigour as underperformance is unnatural and worth scheduling rather than intending, since nothing about a good stretch prompts anyone to go looking for its cause.

Which one is authoritative

Neither. Their relationship is the reading.

A natural question is which instrument wins when they conflict, and the answer is that the question is malformed. The benchmark is not a claim about what is happening and expectancy is not a claim about what should. There is nothing to arbitrate — the gap between them is itself the output, and it is the quantity the alpha definition is built on. Asking which is right is a category error that usually resolves into believing whichever number is more comfortable.

Why this pairing is unusual

Most systems have one half and call it complete.

Almost every trading approach carries one of these instruments. Systematic approaches typically have the backtest or simulation and no continuous live grade in R. Discretionary approaches typically have some sense of current performance and no distribution to place it in. Each half alone is coherent, useful, and structurally unable to answer whether live behaviour is consistent with expectation — which is the question the whole governance layer downstream depends on being answerable.

Connected inside MARS

TEX — Expectancy →
MON — Monte Carlo →
Live-Trade Benchmark Comparison →
The Comparison Stack →

Every brief documents the same shipped system.

The complete MARS package — eleven workbooks, three TradingView indicators, the full manual library — $497.

Get MARS Pro — $497Explore the Platform