- Jul 20
- 6 min read
The Inelastic Markets Hypothesis argues that order flow, not news, is the primary driver of price: mandate-constrained institutional demand moves markets, and that price impact is exaggerated when there are no easy substitutes.
It has been hard to test directly, because the active risk-taking flow that does the moving is rarely observable — and the flow on the other side, the liquidity that absorbs it, even less so.
Futures are a fertile place to look. Two opposing investor groups dominate: the speculative Managed Money group (systematic, largely trend-following CTAs) that demands liquidity, and the commercial Producer/Merchant group (physical hedgers) that supplies it.
Exponential Technology (XTech) predicts the realized daily positioning of both, and estimates each group's latent reaction curve — so we can measure not one footprint but the interaction that sets the price.
Working from that resulting XTech Futures Flow, six findings stand out.
These 6 findings are described in detail in the white paper "Measuring Market Impact and Inelasticity from Managed-Money and Producer Flow Forecast Data"
The two flows are opposite-signed mirrors
XTech's standardized managed-money flow predictions co-move with the same-day return at +0.37, uniform across sectors; producer flow leans the other way at −0.17, negative in every commodity sector.
On a day speculators are net buyers the price tends to be up; on a day hedgers are net buyers it tends to be down, because they add shorts into rallies and lift them into declines.
Public volume (−0.01) and open-interest change (+0.03) see neither.
The two groups are opposite-signed counterparties in the same trade — the fact everything else rests on.
XTech's signed managed-money flow tracks the same-day return at +0.37 across the universe, while producer flow mirrors it at −0.17; the unsigned public series — volume (−0.01) and open-interest change (+0.03) — carry essentially none of it. The information lives in the sign of the flow. Positions-table signed flow vs same-day return; managed money on 82 markets, producer on 38 commodity markets; 2011–2026.
Their market impact multipliers are opposite-signed too, and both persist.
Estimated per market (Gabaix–Koijen, walk-forward), the multiplier is
positive for managed money — largest in thin, concentrated markets (energy, single commodities)
near zero in deeply liquid rates and FX, recovering the inelasticity ordering exactly as theory predicts
negative for producers in every commodity sector, the hedging signature of fading the move.
Unlike mandate-driven equity flow, whose impact decays with horizon, neither futures series collapses toward zero: the signs, and most of the magnitudes, persist with little to none temporary market impact.
Stripping out volatility also surfaces hidden inelasticity on the managed-money side — Class III Milk sits 58 rank-places above where its volatility alone would place it — a market-level risk signal invisible in price laid bare by XTech’s Futures Flow Forecast.
Estimated per market, the Gabaix–Koijen multiplier is positive for XTech's managed-money flow — thin markets (energy, single commodities) move most per unit of flow, deep markets (rates, FX, treasuries) least — and negative for producers, who fade the move in every commodity sector. Median walk-forward daily multiplier M by sector, both groups; 2011–2026.
The two reaction curves are a stabilizing book and a momentum book
Read directly off XTech's daily updated scenarios tables, the producer book is the classic stabilizing latent order book Jean-Pierre Bouchaud's latent liquidity theory posits, but the equity study could only assume — buy below the price, sell above — in 36 of 38 markets.
The managed-money book is its opposite, trend-following in all 38: sell below, buy above, amplifying moves.
Yet cross-market market impact is priced by the demander's curve (managed money, rank −0.84 between nearby flow and the multiplier), not the stabilizing supplier's (−0.09) — a clean empirical separation only visible with both books in hand.
Read off XTech's scenarios graphs: producers quote the stabilizing shape (buy dips, sell rallies) in 36 of 38 markets; managed money quotes the trend-following shape (sell dips, buy rallies) in all 38. Cross-market impact is priced by the liquidity taker’s curve, not the supplier's. Mean reaction to reference price shocks, both books; 38 commodity markets, 2011–2026.
The interaction tells you which moves last
The two XTech model flows are negative mirrors (daily correlation −0.10, deepening to −0.51 on position levels).
At daily resolution managed-money flow co-moves with the return the same day and only the same day, while producer flow responds over the following few days and never leads.
And when producers absorb a large speculative move it reverts, while a move they do not absorb continues — roughly a half-percentage-point difference in the 20-day forward return. Hedger participation is an observable marker of which moves subsequently give back.
Forward return after large managed-money-flow days: when producers absorbed the move it reverts; when they did not, it continues — about half a percentage point over 20 days. Hedger participation marks which moves give back. Signed in the day's own direction, flow-size matched; 2011–2026.
Both models foretell the CFTC ground truth
Against the weekly Commitments of Traders report — the indubitable ground truth public record of real positioning — XTech's forecasted managed-money direction matches real Managed Money about 71% of the time across the commodity markets, every market beating chance by a wide statistical margin (the bucket is a superset of the CTAs modeled, so agreement is strong but imperfect).
Both add little in deeply liquid financials, exactly where market impact is near zero — an expected limitation.
External validation against the CFTC COT report: XTech's forecasted managed-money direction matches real Managed Money ~71% of the time, and the producer model calls its exact Producer/Merchant category ~74% — every market above chance, with the forecast in hand three full trading sessions before publication. Weekly directional agreement, each model vs its matching CFTC group; common markets, 2011–2026.
Each group supports a distinct tradeable strategy
A producer release-cycle strategy — forecast the weekly hedger report before publication, trade the drift around it — earned a net Sharpe of 1.24 (2.28 Sortino) after realistic execution costs, 7.1% a year at 5.7% volatility with a 6% worst drawdown, and strengthened in the recent subperiod.
A separate managed-money conviction signal, built from two XTech Futures Flow positioning fields with no price input, earned a net Sharpe of 0.82 market-neutral, roughly tripling a dollar at a −12% drawdown against −31% for buy-and-hold.
Its forward rank information coefficient is 0.05 (t-statistic = 5.9, non-overlapping windows): highly significant, and a clear source of the edge.
Two low-correlation books from one XTech dataset. Just two of myriad strategies.
The producer release-cycle strategy — position for the hedger report before publication, refine on the released number — net of realistic costs: 7.1% a year at 5.7% volatility, net Sharpe 1.24, with a 6% worst drawdown. Growth of $1 compounded, net of costs, vs buy-and-hold; 27 commodity markets, 2014–2026.
The managed-money conviction strategy, built from XTech positioning data alone: net Sharpe 0.82 market-neutral, roughly tripling a dollar at a −12% drawdown versus −31% for buy-and-hold. Growth of $1, each series at its own realised volatility, 2011–2026.
Taken together, the direction of each group's XTech-predicted flow, the structure of its reaction curve, and the interaction between the liquidity demander and supplier of liquidity carry information that price and volume data do not — contemporaneously, as market impact, and forward, as tradeable signal.
Data Source: XTech Global Futures Flow Forecasts
The analysis draws on XTech Global Futures Flow Forecasts, which models both trader groups. The full methodology, figures, and results are in the white paper, "Measuring Market Impact and Inelasticity from Managed-Money and Producer Flow Forecast Data."
Comments