Commitments of Traders: Strategies for Tracking the Market and Trading Profitably
Floyd Upperman · 2006 · 222 pages
Upperman turns the weekly CFTC Commitment of Traders report into a leading indicator by tracking how far commercial hedgers and funds sit from their own historical extremes. The idea is sound and still used today. The actual math that decides those extremes is never shown.
The one idea
The book's core claim is that the CFTC's weekly Commitment of Traders (COT) report, broken into commercial hedgers, funds (noncommercials), and small speculators, is a leading indicator of price because commercial producers and consumers are closer to real supply and demand than anyone trading off a chart. Upperman calls this Individual Market Participant Analysis (IMPA): watch the net-commercial position for statistically extreme readings, treat those as an early warning that a market is out of balance, and only then bring in lagging price tools to time the actual trade.
The mechanism is the UCL/LCL indicator: upper and lower commercial limit lines built from a weighted historical mean and a proprietary number of standard deviations. When the net-commercial position crosses one of those lines, it flags a statistically rare event (Upperman notes a plain 3-standard-deviation boundary would only be crossed 0.27 percent of the time, and his wider EUCL/ELCL extreme lines are calibrated to roughly that rarity). That alone is not a trade. The system also requires the opposing side, usually the funds, to be holding an extreme position of its own, a proprietary RSI diverging in the direction of the setup, and price structure confirming a new trend, most often a close above or below the 18-day moving average for two straight days.
Once in a trade, the book is explicit about mechanics: stops are set on closing prices only (never intraday), profits are taken in halves under the '50 percent rule,' and position sizing is capped by a 10 percent rule (dropping to 5 to 8 percent above $100,000 in the account). The book backs the UCL/LCL trigger with a 1998 study across 38 markets, paired with simple 10- and 18-day moving averages for entries and exits, that Upperman says produced a net profit in every market tested.
What it actually teaches
The IMPA setup trade runs through seven steps, in order, from spotting the COT anomaly to closing the position.
READ THE COMMERCIAL GRAPHS
Check the net-commercial position against its UCL/LCL lines every week when new COT data drops. A position at or beyond the line means commercial hedgers are holding a statistically one-sided stance versus their own multi-year history. This is the trigger; nothing else in the system starts without it.
CHECK RSI DIVERGENCE
On the daily chart, look for Upperman's proprietary RSI (a recency-weighted variant of the standard formula, plotted directly over price rather than in a separate window) to diverge from price in the direction the COT trigger points. A market making a new high while the RSI fails to make a new high is negative divergence, a sign the move is weakening from the inside.
READ THE RAW TECHNICALS
Layer in the daily price chart: open, high, low, close, the 18-day and 10-day moving averages, and day-of-week bias built from at least 30 samples per weekday (the book cites the S&P 500 closing up 56 percent of Mondays over five years and 65 percent over the trailing 161 sessions as an example of the kind of pattern this step surfaces).
CHECK OPEN INTEREST
In a sideways market, falling open interest is read as bullish (shorts covering); rising open interest in a sideways market is read as bearish (shorts adding). Upperman notes this step mattered more before 2000, when COT data was released biweekly or monthly; weekly releases make open interest tracking a secondary confirmation rather than the primary signal it used to be.
PULL THE SUPPORTING GRAPHS
Split the commercial category into assumed producer (short) and consumer (long) positions and check which side is driving the imbalance. The ideal buy setup: commercial consumers at or near their UCL while commercial producers sit inside their normal range (near center, not at their own LCL), which the book reads as producers unwilling, not unable, to meet demand.
ENTER ON THE 18-DAY BREAK
Buy (or sell) on a close above (or below) the 18-day moving average for two consecutive days, the '18-day moving average two-day rule.' If that entry is missed, wait for price to regress back to the 18-day or 10-day moving average rather than chasing. Every entry requires a logical stop tied to a chart level, price pattern, or moving average, capped at 10 percent of account capital ($2,500 to $99,999 accounts; 8 percent at $100,000, down to 5 percent or less above $1,000,000); if the logical stop implies more risk than the cap allows, the trade is skipped, not resized with a tighter monetary stop.
EXIT IN HALVES
Trade an even number of contracts. Close roughly half the position at a first profit objective (the book's own soybean-meal example took half off after a $1,250 gain against a $940 initial margin), move the stop on the remainder to breakeven or better, and trail the rest with the 18-day moving average once a trend is confirmed (10-day average crossing the 18-day, both sloping with price). Stops are read on closes only, never touched intraday, and are never widened past their original level.
What it looks like Monday morning
A reader with a futures account and charting software that can plot the CFTC's weekly COT data (free from cftc.gov) can rebuild the outer layer of this system on Monday: pull net-commercial and net-fund positions for a handful of markets, plot them against a multi-year history, and flag any market where the current reading looks like an outlier against its own past. That much is fully reproducible from the book.
What a reader cannot rebuild is the actual trigger. The UCL/LCL lines, the weighting scheme for the recency-weighted RSI, and the exact number of standard deviations used for the 'extreme' EUCL/ELCL lines are all called proprietary and never given as a formula. So Monday's real task is narrower than the book implies: track raw net-commercial and net-fund extremes by eye or with a plain standard-deviation band, then layer in the book's fully disclosed mechanics, the 18-day two-day entry rule, closing-only stops, the 50 percent rule, and the 10 percent risk cap, on top of that home-built approximation.
| Element | Forward plunger (bullish) | Reverse plunger (bearish) |
|---|---|---|
| Formation | Makes a new 10-day low, then closes in the top 30% of that day's range | Makes a new 10-day high, then closes in the bottom 30% of that day's range |
| Minimum range | Day's full range must be at least one-third of the market's average daily range | Same |
| Follow-through window | Must close above the plunger-day high within 3 days or it expires | Must close below the plunger-day low within 3 days or it expires |
| Invalidation | Plunger-day low broken on a close during the 3-day window | Plunger-day high broken on a close during the 3-day window |
| Logical stop | Plunger-day low | Plunger-day high |
A plunger that fails immediately is treated as a signal in the opposite direction, not just a dropped setup.
The market does not care about how much money you have to risk.Upperman, Commitments of Traders
Where it fails
- The core indicator is never shown. The UCL/LCL calculation, described repeatedly as 'proprietary,' is the single input every other step of the IMPA system depends on, and the book gives only its logic (a weighted mean plus an undisclosed number of standard deviations), not the formula. A reader following this edition literally cannot reproduce the trigger that starts every trade; they can only approximate it with a plain standard-deviation band and hope it behaves the same way.
- The 38-market backtest cannot be checked. The book's central proof point, a 1998 study claiming a net profit in all 38 markets tested when the UCL/LCL trigger was paired with 10- and 18-day moving averages, is presented as a hypothetical computer test, not real trades ('the trades were not actually executed'). Worse, the appendix pages that were supposed to hold the per-market results (pp. 154-167 in this printing) extracted as pages of table borders and headers with no readable data, so even the summary claim can't be audited from the book itself.
- Commercial versus speculative is an assumption, not a fact. The whole 'producer versus consumer' split inside the commercial category, which drives Step 5, is explicitly a guess: the CFTC report does not break commercials down that way, and Upperman states plainly that a commercial can hold a speculative account, that account type is self-selected at setup, and that 'cross-contamination' between categories is likely. The book asks the reader to trust an inference layered on top of an already-undisclosed statistical trigger.
- The book flags its own obsolescence risk and doesn't resolve it. Upperman writes, while the book is going to press, that commodity index funds are increasingly showing up inside the 'commercial' category even though they are not classic hedgers, which would corrupt the net-commercial signal he built the whole system on. He says he is 'still considering' how to handle this and offers no fix, just a warning that the purity his core trigger depends on may already be eroding as of 2006.
- This is a full-time, well-capitalized trader's book. The system requires daily custom charting across dozens of markets (commercial, consumer, producer, fund, RSI, open interest, equity, and weekly graphs, each updated and cross-checked), multi-week to multi-month holding periods, and futures margin plus the 10 percent risk-per-trade buffer just to take one position. Nothing here is built for someone checking a phone once a day or trading a small account close to the $2,500 floor the risk table starts at.
- Worked examples are all winners. Chapter 9's trade walkthroughs (the 2005 hog buy, the Russell IMPA buys, the gold buy) are all successful setups presented after the fact with the benefit of knowing which one worked. The book states elsewhere that not every UCL/LCL penetration leads to a real turn and that 'nothing works 100 percent of the time,' but no losing IMPA trade is walked through step by step to show what a failed setup looks like in real time.
Who it is for
Buy it if
You already trade futures with real capital, can build or license custom COT charting, and want a structured way to combine CFTC positioning data with standard trend tools like moving averages and RSI. It's also worth reading for the risk mechanics alone: the closing-only stop, the 50 percent scale-out rule, and the tiered 10 percent risk cap are clearly stated and usable on their own, independent of the IMPA trigger.
Skip it if
You want a reproducible, back-testable system with disclosed math, or you trade stocks and expected the COT angle to translate directly (Upperman's own framing is that commodities behave differently from equities, mean-reverting where stocks don't). Skip it too if you don't have the screen time or capital the multi-graph, multi-week IMPA process assumes.
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.