Williams sells a stack of day-of-week filtered breakout and reversal patterns tested on 1990s bond and S&P futures data, then spends his best chapter on position sizing. The patterns are dated curve-fits; the position-sizing math is the part worth stealing.
The one idea
Williams' short-term method rests on a claim that markets are not random: prices spurt into one-to-five-day trend explosions driven by news, crowd emotion, and the relationship between related markets like bonds and stocks, and a trader's job is to identify the setup, time the entry into the explosion, and manage the trade once in it.
His entries fall into two families. The first is the volatility breakout: add a percentage of the prior day's range to tomorrow's opening price and buy (or subtract and sell) if price reaches it, on the logic that a market strong enough to extend past a cushion above the open is showing real buying pressure, not noise. The second family is a set of pattern-based reversal setups (Oops!, Smash Day, the Specialists' Trap) that all share the same skeleton: the public gets emotionally sucked into a breakout, the breakout fails the very next day, and that failure becomes the entry signal in the opposite direction.
Every setup in the book is then run through the same filters: Trade Day of the Week (some days trade far better than others), a bond/stock intermarket bias, and a short moving-average overbought/oversold reading. The last quarter of the book abandons chart patterns entirely for money management, arguing that pattern selection barely matters next to how many contracts you put behind each trade.
What it actually teaches
Williams builds a trade in four stages, matching his own selection, timing, and trade-management framework plus a bolted-on sizing rule.
SELECT WITH BACKGROUND FILTERS
Before looking at any entry pattern, apply the context filters: Trade Day of the Week (his bond tests found Tuesday and Thursday the best buy days, Wednesday and Thursday the best sell days), an intermarket bias (buy S&P futures only when bonds closed higher than 15 days ago, sell only when lower), and a short-term overbought/oversold read off a 9-day moving average. He treats these as mandatory, not optional: unfiltered systems in his own tables show far worse average profit per trade and much larger drawdowns than the same system with filters applied.
TIME THE ENTRY
Two mechanical entries carry the book. The volatility breakout adds a percentage of the previous day's range to tomorrow's opening price (his tests favored buying 100 percent of the prior range above the open on bonds, and separately a 40/200 percent split on the S&P). The Oops! pattern buys when the market opens below the previous day's low, then rallies back up through that low the same day (and mirrors for a sell above the prior high); he calls it "the single best pattern I have discovered" and dates his own use of it back to 1978.
SIZE THE POSITION
Williams rejects the Kelly formula he traded through the 1980s (F = ((R+1)*P-1)/R) after it produced huge, uncontrolled equity swings, and replaces it with his own rule: (account balance x risk percent) / largest loss in the system = number of contracts to trade. He suggests 5 percent for a cautious trader up to 20 percent-plus for an aggressive one, and states that risk above roughly 21 to 25 percent makes drawdown grow faster than profit.
EXIT WITH THE BAILOUT RULE
Most examples in the book exit on a flat dollar stop (commonly $1,500 to $2,500 depending on the market) paired with what he calls the bailout exit: close the trade on the first profitable opening after entry, rather than holding for a bigger target. Several of his best-performing tables extend that to a fixed multi-day hold (2 to 4 days) before applying the bailout rule.
What it looks like Monday morning
A reader who takes this seriously does not copy the day-of-week table into a live system. What transfers is the habit of testing entries against a background filter (which days, which direction bonds are leaning, whether the short-term average is overbought) rather than trading a pattern in isolation, and of writing down, in advance, exactly what percentage of the account backs each trade before a losing streak forces the decision under stress.
The concrete change worth making is switching from a fixed contract count or an arbitrary stop size to Williams' own formula: account balance times a chosen risk percentage, divided by the largest loss the system has produced. It is crude next to what a modern quant desk runs, but it is a specific, testable number a discretionary trader can actually compute before Monday's open, which almost none of the pattern material in the rest of the book offers.
| Approach | Rule | Result |
|---|---|---|
| No money management | 1 contract on every trade | $251,813 |
| Ryan Jones Fixed Ratio | add a contract on a fixed ratio tied to largest drawdown x 2 | $18,107,546 |
| Williams' risk-percent formula | (account balance x 15%) / largest loss = contracts | $582,930,624 |
These are backtested paper results scaling a single bond system up to as many as 5,000 contracts on a $20,000 starting balance, not live trading records, and Williams says as much in the surrounding text.
I believe the current trade I am in will be a loser ... a big loser at that.Larry Williams, Long-Term Secrets to Short-Term Trading
Where it fails
- The patterns are curve-fit to one decade of a market structure that no longer exists. Every backtest in the book runs on data ending in July or August 1998, and several results are built on small samples: 25 occurrences of one S&P pattern, 41 or 57 trades on a bond outside-day setup. Williams himself notes that Bonds "traded totally different" after the exchange added overnight sessions in October 1988, and that a Thursday Fed-report timing quirk that drove his Friday bond edge disappeared by the time he wrote the book. The book documents its own edge decaying mid-write and offers no evidence the day-of-week filters survived into the 2000s, let alone the electronic, high-frequency futures markets of today.
- The money-management chapter's headline example is a near-blowup, not a success story. Williams describes trading his own account from $10,000 to $2,100,000 in 1987 using the Kelly formula, then watching it collapse back down to $700,000 before recovering to $1,100,000 by year end, a roughly 67 percent drawdown from the peak. He presents this as evidence Kelly is flawed and his own risk-percent formula is the fix, but the paper result he then shows for that fix, $582,930,624 compounded from a $20,000 account trading up to 5,000 bond contracts, is explicitly a backtest, not a trading record, and the book does not walk through what a comparable real-money drawdown on his own formula would look like.
- The book assumes testing infrastructure most readers do not have. Every chart and table cites ULTIMANAGER (money-management simulation software) and the Genesis Navigator charting platform, both commercial tools from the late 1990s. A reader cannot verify a single one of the book's headline numbers, the 180 percent Greatest Swing Value multiplier, the $573 average profit per trade, without re-running the same backtests on current data using different software, which the book gives no guidance on doing.
- Every walked-through example is a winner; none of the losing variants are shown. The book repeatedly says things like "can we do better?" and then shows an improved version of a system, but never shows the discarded alternatives that did not test well. Given how many free variables are available (which day of week, which moving-average length, which intermarket filter, which multiplier on the swing value), the presented results read as the survivors of an unshown optimization process rather than an out-of-sample test.
- It is a futures book wearing a general trading title. Nearly every worked example is bonds or S&P 500 futures, with a handful of commodities (Cattle, Copper, Silver, Wheat). A reader who trades only stocks gets almost nothing directly usable; the volatility-breakout and Oops! mechanics translate conceptually, but none of the specific thresholds were tested on equities in this edition.
Who it is for
Buy it if
You already have futures backtesting software and 1990s-onward bond and S&P data, and you want to see one trader's full worked example of turning a pattern idea into a filtered, sized, and exited system, especially the position-sizing chapter and its comparison of Kelly, Ryan Jones' fixed-ratio, and Williams' own risk-percent formula on the same data.
Skip it if
You want a system to run this week without re-verifying its numbers, you trade only stocks, or you are looking for a book that spends serious time on why systems fail rather than showcasing the ones that worked. The autobiographical chapters (roughly the first third of the book) also carry little tradeable content if you are here strictly for the method.
As an Amazon Associate I earn from qualifying purchases. It costs you nothing extra, and it does not change what gets recommended or what the verdict says.