The Almanac · Vol IGlossary35 entries

The glossary

The vocabulary the Investing Engine uses, plus the general trading terms it assumes — defined plainly, cross-referenced, and linked back to where they appear in the published results.

A2 entries

All-Time High (ATH)

The highest price an instrument has ever traded at. Retracement levels are commonly measured down from a running ATH, which is why the figure matters even long after it was set.

Allocation Depth

How much of an intended position to commit, set by which retracement level price is currently trading at — the deeper the level, the larger the commitment. This is the sizing half of a decision; the other instruments answer the timing half.

B2 entries

Backtest

Running a model over historical price data to see what it would have signalled and what that would have returned. A backtest is evidence about the past, not a prediction. Three of the five instruments render one live on any chart you load them onto.

Buy-and-Hold

Buying at the start of a test window and holding to the end, with no exits. The benchmark every published result is measured against. A system that takes exits will underperform buy-and-hold during a sustained one-directional advance — that is arithmetic, not a defect.

C2 entries

Configuration

Setting a model's inputs to suit the asset class, the horizon and the thesis you hold. Legitimate and required — the model executes a thesis, it does not predict direction. What separates it from curve fitting is a rule stated before the result is seen.

Confirmation

A second, independent reading that agrees with the first. Confirmation reduces the chance that a signal is an artefact of the one thing that produced it — which is the entire argument for using the instruments together rather than alone.

D2 entries

Divergence

Price and an indicator moving in opposite directions — price making a higher high while the indicator makes a lower high, or the reverse. Commonly read as a weakening of the move in progress.

Drawdown

The fall from a peak in account value to the lowest point that follows, before a new peak is made. Max drawdown is the deepest such fall across a test. It describes what you had to sit through, which return alone never does.

E2 entries

Equity Curve

Account value plotted over the life of a test. Its shape carries information the final number hides — two systems can end in the same place having delivered completely different experiences along the way.

Exposure

The share of bars in a test during which a position was open. The remainder was spent in cash and therefore not at risk. Comparable return at lower exposure means the same participation for less time exposed to a fall.

F1 entry

Fibonacci Retracement

Horizontal levels drawn at fixed proportions of a prior move — commonly 38.2%, 50% and 61.8% — used to mark where a pullback might find support. A long-established, widely published technique.

I1 entry

Instrument

One model within the stack: Convergence, Deviation, Trend, Adaptive Trend, or Peak Retracement. Never sold, priced or licensed separately — a subset cannot produce the decision the five produce together.

RelatedThe Stack
M4 entries

Mean Reversion

The tendency of a price stretched far from its recent average to return toward it. The counterpart to momentum, and the reading that fails hardest when a stretch simply keeps stretching.

Model Setting

An input that changes which signals fire — directional bias, tightness, trading style, noise suppression. Because it changes the output, it needs a rule stated before the result is seen.

Momentum

The rate at which price is moving, and whether that rate is building or fading. Momentum readings tend to turn before price does, which is what makes them useful and what makes them prone to early signals.

Moving Average

The average price over a trailing number of bars, recalculated each bar. Smooths noise at the cost of lag — a longer average is steadier and slower, a shorter one is faster and noisier.

N1 entry

Noise Suppression

A model setting that filters out small deviations so only larger ones register. Higher values mean fewer signals; zero reports everything the model sees.

O2 entries

Oscillator

An indicator that moves within a bounded range rather than tracking price directly. Used to judge whether a move is stretched relative to its own recent history.

Overbought / Oversold

A reading that price has moved unusually far in one direction relative to its recent range. Neither is a signal on its own: a market can stay stretched far longer than a position can be held against it.

P2 entries

Portfolio Setting

An input that changes the accounting but not the signals — entry percentage, exit type, starting capital, fees. Disclosed rather than justified, because varying it cannot curve-fit the model.

Position Sizing

Deciding how much capital to commit to a trade. Sizing determines what a wrong call costs, which makes it at least as consequential as the entry itself.

R2 entries

Regime

A stretch of market behaviour with consistent character — trending, range-bound, high or low volatility. A model that performs well in one regime can perform badly in another, which is why results are sliced by regime rather than averaged.

Risk-Reward Ratio

The size of the intended gain against the size of the loss accepted if the trade fails. A favourable ratio can carry a low win rate; an unfavourable one can lose money at a high win rate.

S3 entries

Slippage

The gap between the price a trade was expected to fill at and the price it actually filled at. Published backtests set slippage to zero, and that is disclosed on every result — real fills will be worse.

Stop Loss

A predetermined exit price that caps the loss on a position. Setting it before entry is what makes it a rule rather than a reaction.

Support and Resistance

Price areas where a move has repeatedly stalled or reversed — support below, resistance above. Levels, not lines: they mark zones where the balance of buyers and sellers has previously shifted.

T5 entries

The Stack

Using the instruments together so their signals confirm one another — a stretch reading, a reversal reading and a confirmed trend turn agreeing is the high-conviction state, and a retracement level then sets how deep to commit. The stack is the product; the individual instruments are its components.

Tightness

A model setting governing how closely a trend read tracks price. Tighter settings react sooner and produce more signals; looser settings react later and produce fewer, which suits longer horizons.

Timeframe

The period each bar on a chart represents — 4-hour, daily, weekly. The same model on the same asset produces different signals on different timeframes, so the timeframe is part of the configuration, not a viewing preference.

Trend

A sustained directional bias in price. Easy to identify after the fact and hard to identify at the turn, which is where confirmation from an independent reading earns its keep.

Tuning

Adjusting inputs by trying combinations until historical return improves — curve fitting. The test that separates it from configuration: if the only justification for a setting is that it backtested best, it is tuning, even when the value chosen happens to match what a rule would have given.

U1 entry

Underwater

The stretch of time spent below a previous peak in account value, before that peak is regained. Depth is measured by drawdown; duration is measured here, and duration is usually what ends people's patience.

V1 entry

Volatility

How much price moves over a given period, in either direction. Rising volatility widens the range of outcomes for a position of unchanged size, which is why sizing and volatility belong in the same decision.

W2 entries

Whipsaw

A signal that fires and immediately reverses, exiting at a loss just before price resumes its original direction. The characteristic cost of reacting quickly, and the reason noise suppression and tightness exist as settings.

Win Rate

The share of closed trades that finished in profit. Reported by the built-in backtest engine, but never led with here: it can be reshaped almost at will by changing when positions are closed, and it says nothing about the size of the losses.

See these terms at work.

Every definition here maps to something a published backtest reports. Read the test protocol to see how each figure was produced.