Risk and position sizing: level-based stops change distance on every trade, so size must change too
Module 13 put a stop at a price that the leg justified. This module makes sure that stop risks the same money every time, because the distance from entry to a level-based stop changes with every leg and every entry level, and a trader who keeps the lot size constant is risking a random amount on each trade. The method is fixed fractional risk: decide the money you will lose if the stop is hit, as a fraction of the account, and derive the position size from that figure and the stop distance. On gold the arithmetic is simple once you know what a lot is worth per dollar of movement. The module then covers the things that quietly break the arithmetic, the allowance around a level, slippage at releases, nested positions that share one risk, and the loss limit that stops you trading a day that has turned into a range. It ends with an honest paragraph on expectancy, with illustrative numbers that are clearly labelled as such.
Fixed fractional risk
Decide, before anything else, the fraction of the account that one losing trade may cost. Half of one percent to one percent is the usual range for a method with level-based stops, because the stops are hit often and the winners are expected to be a few multiples of the risk rather than many. On a 10,000 account at one percent, the money at risk is 100 per trade; at half a percent it is 50. This figure is fixed, and everything else, the lot size in particular, is derived from it. The point of expressing risk as a fraction rather than a lot size is that it survives the two things that change constantly: the account balance and the stop distance. A trader who always trades 0.05 lots is not managing risk; they are letting the stop distance decide it for them. The fraction should also be one you can lose several times in a row without changing how you trade, because level-based stops produce losing streaks as a matter of course, and a trader who tightens up after three losses has stopped running the method.
The sizing formula on gold
One standard lot of XAUUSD is 100 ounces, so a one-dollar move in the gold price is worth 100 in account currency per lot, and 0.01 lots, the smallest size most brokers allow, is worth one per dollar of movement. The position size in lots is therefore the money at risk divided by the stop distance in dollars, divided by 100. For the 61.8 entry in Module 13, entry 3076 and stop 3034, the stop distance is 42 dollars; with 100 at risk the size is 100 divided by 4200, which is 0.0238 lots, rounded down to 0.02. For the 38.2 entry, entry 3124 and stop 3068, the distance is 56 dollars and the size is 0.0178, rounded down to 0.01. Always round down; rounding up risks more than the plan. Brokers quoting gold in other units or with different contract sizes change the constant, not the method, so check the contract specification once and write the constant into your plan.
Why size changes every trade
The two examples above risked the same 100 and used different sizes because the stops were different distances. That is the whole point, and it is where fixed-lot traders go wrong. A trader who used 0.02 lots on both would have risked 84 on the first and 112 on the second, and on a day when the level is a wide H4 zone with a 90-dollar stop, 180. The variation is not noticed trade by trade and it ruins the journal, because the results no longer reflect the method but the accident of which stops were wide. Every entry in the Module 15 journal records the stop distance and the size, and the money at risk should be the same figure on every line. If it is not, the first thing to fix is not the levels but the sizing. A simple spreadsheet or calculator that takes the stop distance and returns the lot size removes the arithmetic from the moment of entry, which is where it is most likely to be skipped.
Deep and shallow entries at the same risk
Sizing by stop distance also makes the comparison between entry levels honest. The deep 61.8 entry at 0.02 lots gains 248 at the 161.8 target, which is 2.48 times the risk after rounding, and the shallow 38.2 entry at 0.01 lots gains 200 at the same target, or two times. The deeper entry pays more at the same target for the same money at risk, which is the arithmetic reason the deep zone is favoured, and the shallow entry's advantage is only that it exists on days when the deep zone is never reached. Neither is better in the abstract. What the sizing rule ensures is that whichever you take, a loss costs the same, so the journal can compare them on their results rather than on their stop widths. The same comparison applies across timeframes: an M15 entry inside an H4 zone has a stop of perhaps twelve dollars and a size several times larger than the H4 limit, and at the same money risk its targets in multiples are correspondingly larger.
The allowance, spread and slippage
Module 3 described a level as a band with an allowance either side, and the stop in Module 13 sat beyond the next level plus that allowance. The allowance costs stop distance, and the temptation is to remove it to improve the arithmetic on paper. Resist it; the correct response to a wide stop is a smaller size, never a stop inside the zone. Add the spread, which on gold is a dollar or two in normal conditions and several at the open or around a release, to the stop distance before sizing, because the stop fills at the bid or ask rather than at the chart price. Around scheduled releases, assume slippage of ten to twenty dollars beyond the stop and either flatten before the release or size as if the stop were that much further away. A stop that fills twenty dollars past its price on a 42-dollar stop has cost half as much again as planned, and that is the kind of loss that makes a method look worse than it is.
Correlated and nested positions
Multi-timeframe entries from Module 8 create a trap. A limit at the H4 61.8 and a confirmation entry on the M15 leg inside the same zone are two orders, but they are one idea with one invalidation, and if the zone fails both are lost together. Size them as one position against one risk figure: if the plan allows 100 at risk on the idea, split it between the two entries rather than giving each its own 100. The same applies to a position added on a retest, to a second gold position on a different timeframe in the same direction, and to any pair of trades that would be stopped by the same move. The question to ask is whether one candle could take both stops; if it could, they are one trade for sizing purposes. The reverse case, a long on one timeframe and a short on another, is not a hedge but two trades with two risks, and it usually means one of the two timeframe reads is wrong and should be resolved before either position is held.
Daily and weekly loss limits
Set a loss limit for the day, commonly two or three times the per-trade risk, and stop when it is reached. On a Fibonacci method the limit does double duty. Three level failures in a session are not just three losses; they are evidence that the session is a range or a chop in which the tool has nothing to measure, which Module 6 said to stand aside from. The limit forces that decision when the trader would rather keep drawing. A weekly limit of roughly three daily limits does the same for a week in which gold is between trends. Neither limit is a judgement on the method; both are a recognition that a tool for trends should be put away when there is no trend, and that the account should still be intact when the next one starts. Write the limit as a number of trades rather than a sum of money, because it is easier to count three stops than to track a running total while the chart is still open.
Expectancy, with honest numbers
Expectancy is the average result per trade in units of risk: the win rate multiplied by the average winner, less the loss rate multiplied by the average loser. These are illustrative numbers and not a measured result for any level or method: if four trades in ten won an average of 2.5 times the risk and six lost one times the risk, the expectancy would be 0.4 units of risk per trade, which at 100 of risk is 40 per trade on average over a long series. The same arithmetic with three wins in ten and 2.5 average winners gives 0.05, barely positive, and with 2.0 average winners it is negative. The lesson is not the numbers but the shape: with level-based stops the win rate will be modest, so the method lives or dies on the size of the winners relative to the losers, which is why the target ladder and the trailing rules matter as much as the entry. Your own figures come only from the journal, after enough trades to mean something.
FAQ
How much should I risk per trade with Fibonacci stops?
A fixed fraction of the account, usually half of one percent to one percent, decided before anything else. The lot size is then derived from that money figure and the stop distance, so that every trade loses the same amount if the stop is hit regardless of how wide the level-based stop is.
How do I calculate position size on XAUUSD?
Divide the money at risk by the stop distance in dollars, then by 100, because one lot is 100 ounces and a one-dollar move is worth 100 per lot. For 100 at risk and a 42-dollar stop, 100 divided by 4200 gives 0.0238 lots, rounded down to 0.02. Check your broker's contract size once and write the constant into your plan.
Why not just use the same lot size on every trade?
Because the stop distance changes with every leg and entry level, so a fixed lot size risks a different amount each time: 84 on a 42-dollar stop and 180 on a 90-dollar stop at 0.02 lots. That variation makes the journal meaningless, since results then reflect stop widths rather than the method.
Should I count an H4 limit and an M15 entry in the same zone as two trades?
No. They share one invalidation, so if the zone fails both are lost together. Size them as one position against one risk figure, split between the two entries. The test is whether a single candle could take both stops; if it could, they are one trade.
What is a reasonable expectancy for a Fibonacci method?
No figure can be given honestly, because it depends on your swings, your confirmation rules and your targets. The illustrative shape is that with level-based stops the win rate is modest, so the method depends on winners being a few multiples of the losers. Your own expectancy comes from your journal after enough trades to mean something.