Expectancy is the average R-multiple of your trading system: the mean of all your trades' results when each is expressed as a multiple of its initial risk. A system with an expectancy of 0.8R earns, on average, 0.8 times the risked amount per trade, so over 100 trades it would tend to produce around 80R. Van Tharp made expectancy the headline statistic of system analysis because it captures what win rate hides: a system can lose on most trades and still be strongly profitable, or win on most trades and still be a slow bankruptcy. This is part 6 of Artha's 10-part series on Tharp's Super Trader (McGraw-Hill, 2009); the concepts are his, the worked examples are ours, and this is education, not advice. New readers should start with R-multiples in part 5, since expectancy is built directly on them.
Chapter 1What is expectancy, precisely?
Expectancy is the sum of all your trades' R-multiples divided by the number of trades: the mean R per trade. If your last ten trades produced +4R, -1R, -1R, +2.5R, -0.7R, +0.9R, -1R, -1.7R, +5R, and -1R, the total is +6R across 10 trades, so the expectancy is +0.6R. On average, each trade in this sample earned 0.6 times its initial risk, despite six of the ten being losers.
Because expectancy is denominated in R, it says nothing about rupees until you attach a position size. That separation is deliberate and is the hinge of Tharp's whole framework: the system's job is to produce a healthy R distribution, and position sizing's job, covered in part 7, is to convert R into money at a rate that suits your objectives.
Chapter 2How does the marble game prove the point?
Tharp's workshop demonstration puts a trading system inside a bag of 100 marbles. In one version, 20 of the marbles are winners paying +10R each, 70 are losers costing -1R, and 10 are bad losers costing -5R. Draw a marble at random, replace it, repeat: that is trading this system.
Most people recoil from the 80% loss rate. But add up the bag: the winners contribute +200R, the ordinary losers -70R, and the bad losers -50R, a net of +80R across 100 marbles. Expectancy is +0.8R per draw. Play long enough and this "terrible" system grinds out large profits, provided, and this is the entire catch, the player sizes each draw so that the inevitable losing streaks cannot destroy the account. With an 80% loss rate, streaks of eight or ten straight losses are routine, which is why the same bag bankrupts players who bet big and enriches players who bet small.
The other lesson from the game is about people, not marbles: when a room of players trades identical draws, their ending equities scatter enormously, from ruin to riches, because each player sizes differently. Same expectancy, different outcomes. Tharp's conclusion was that the "how much" decision, not the trade selection, drives results, which is why this series gives it a full article next.
Chapter 3Why does win rate mislead traders?
Because the human mind treats being right as the goal, while the account only experiences the product of frequency and magnitude. A system that wins 70% of the time with +0.5R winners and -2R losers has an expectancy of 0.7 × 0.5 minus 0.3 × 2, which is -0.25R: a reliable loser that feels great most days. The 20%-winner marble system feels awful most days and compounds wealth. Tharp regarded the need to be right as one of the most expensive biases in trading, driving traders to book tiny profits quickly and let losses run, the exact opposite of the golden rule of cutting losses short and letting profits run.
What is the quality ratio and the "Holy Grail" system?
Tharp judged a system not by expectancy alone but by the ratio of expectancy to the standard deviation of its R-multiples, a signal-to-noise measure of how smoothly the edge is delivered. His rough guideline bands: below about 0.20 is poor to barely tradable, 0.20 to 0.25 is average, 0.25 to 0.30 is good, 0.30 to 0.50 is excellent, 0.50 to 0.70 is superb, and above 0.70 is what he playfully called Holy Grail territory, systems so smooth that aggressive objectives become achievable.
Two refinements matter. First, opportunity counts: a superb ratio on a system that trades once a year is worth little, while a moderate ratio on a system producing 20 trades a month can be extraordinary, an insight Tharp later formalised in his proprietary System Quality Number (SQN). Second, ratios in the Holy Grail range are realistically achievable only when a system is confined to the market type it was designed for, which links this article back to part 4: a quiet-bull system measured only in quiet bull markets can look superb, and the same system measured across all conditions will not.
Counterintuitively, adding a giant winner can lower system quality: a rare +30R trade raises the standard deviation more than the mean, so the ratio falls. Smooth systems are built from many decent winners with low variability, not from occasional lottery tickets.
Chapter 5How do you estimate your own expectancy?
From your R-multiple journal, exactly as built in part 5: sum the R column and divide by the number of trades, and compute the standard deviation of the same column for the quality ratio. Tharp's minimum sample was around 30 closed trades before the numbers mean much, and even then they remain estimates that drift as market type changes; he recommended recalculating continuously so you always know your system's current state.
A worked rupee illustration (numbers illustrative): a trader risks Rs 5,000 per trade with 1R equal to Rs 5,000. Her last 40 trades sum to +14R, so expectancy is +0.35R, and the standard deviation of her R column is 1.4R, giving a quality ratio of 0.25, decent. Over a 100-trade year at this expectancy she would tend to earn about 35R, roughly Rs 1,75,000 before costs at constant sizing, and materially more if the 1R risked grows with the account, since percent-risk sizing compounds. The same arithmetic warns her what a mistake costs: each -4R lapse of discipline burns more than eleven average trades' worth of edge.
Chapter 6What are the classic misuses of expectancy?
Three appear constantly. First, computing it from too few trades and treating noise as edge: ten trades tell you almost nothing, because one outlier dominates the mean. Second, pooling trades across different market types and different systems into one number, which averages a good system in its element with the same system out of its element and calls the mush "my expectancy." Third, quoting expectancy while ignoring costs: in India, brokerage, STT, exchange charges, GST, and stamp duty are all real, and for high-frequency retail styles they can consume a thin edge entirely, one reason SEBI's F&O studies find the vast majority of individual derivatives traders losing money after costs.
How Nora helps
Nora can compute expectancy and the quality ratio from an R-multiple list you give it, simulate marble-game-style draws so you can feel losing streaks before real money ever does, and explain any statistic in this article with fresh examples, always as education, never as a verdict on what you should trade.
App · coming soonWhat this means for you
Expectancy replaces the question "how often am I right?" with "what does my average unit of risk earn?", and the quality ratio adds "how bumpy is the road?" Together with trade frequency, these numbers let a trader audit any system, including their own record, with the detachment of an accountant. What they cannot do is tell you how much to put on each trade, and that decision, in Tharp's telling the most important one in trading, is where the next article goes: position sizing and the CPR model.
Series credit: this series is based on concepts from Super Trader: Make Consistent Profits in Good and Bad Markets by Van K. Tharp, Ph.D. (McGraw-Hill, 2009). Full credit for the framework belongs to Dr. Tharp and the Van Tharp Institute.