Advanced Hedging Optimization

Three different reasons people hedge

Hedging gets discussed as if it's one thing. It isn't. There are three distinct goals you can have when you hedge an existing bet, and they each call for a different optimal hedge size.

Lock in profit (or loss). You bought a futures ticket at +1000, your team made the final, and you want to sleep well regardless of how the final goes. The full hedge size that makes both outcomes equal is the answer. This is the most common framing and the one most calculators show.

Reduce variance without giving up too much EV. Your bet still has positive expectation but the variance is uncomfortable. You want to scale variance down while keeping most of the upside. This calls for a partial hedge, sized by how much variance reduction you want per unit of EV given up.

Free up bankroll for other opportunities. Your money is tied up in a long-running futures bet and a better opportunity has appeared. Hedging the futures position frees up capital. The relevant math is the EV of the new opportunity vs the EV of letting the existing position ride.

Most articles about hedging conflate these. They aren't the same problem and they don't have the same answer.

The full-hedge math (lock-in baseline)

Setup. You have an existing bet of stake S at decimal odds o. The current line on the opposite side is at decimal odds oh (the hedge price). To equalize payouts in both outcomes:

Worked example. You bet $100 on a NBA team to win the championship at +800 (decimal 9.0). They made the finals. Your team is now +150 underdog (decimal 2.50) for the finals series.

That's the lock-in baseline. The hedge is sized to flatten both outcomes to the same dollar profit. The downside: you've given up all upside variance. Your team winning the championship was worth $900; after hedging, it's worth $360 either way.

The EV-optimal hedge (usually zero)

If your original bet had positive EV at the time you placed it, and the current hedge price is the fair price for the new conditions, then hedging at all is negative EV. Every dollar of hedge stake reduces your expected return because you're paying vig at the hedge price.

The math is straightforward. Let p be the current true probability of your bet winning, computed from the devigged hedge market. Your expected value of letting the original bet ride from this point forward:

Real markets always have vig, so the hedge price is always slightly worse than fair. Hedging at a real market price is always negative EV in the strict mathematical sense, regardless of your existing position. The amount you lose to vig per dollar hedged is the same as the amount you'd lose betting that side fresh at that price.

Implication: pure EV-maximization says never hedge. The bettor who only cares about long-run expected dollars should let every bet ride to resolution and just collect the variance.

The variance-reduction hedge

Most real bettors care about variance, not just EV. The combined position (original bet plus partial hedge) has a variance that depends on how much you hedge:

The variance-EV tradeoff: you can dial the hedge size from 0 (full variance, zero EV cost) to the lock-in size (zero variance, maximum EV cost). Each unit of hedge stake reduces variance and reduces EV in a known proportion. The "right" hedge size depends entirely on how much variance reduction is worth to you per dollar of EV given up.

The Kelly-optimal partial hedge

Kelly says you should size your bets to maximize expected log-bankroll growth. Applied to a hedging decision, Kelly answers: how much should I hedge such that the combined position is at the right size for my current bankroll?

The reasoning. Your existing bet was sized at full Kelly when the original edge was X. The line has moved and your remaining edge from this point is now Y, which is different from X. The Kelly-optimal exposure to the new edge is some fraction of bankroll based on Y. If your current exposure (original ticket value) is larger than the Kelly target for the new edge Y, you should hedge enough to bring exposure down to the right level.

This is the most coherent framework for partial hedging. It tells you exactly how much to hedge based on the current state of your edge, your bankroll, and your risk tolerance encoded in the Kelly fraction (full, half, or quarter).

Worked example: a futures hedge at half Kelly

You bet $200 on a Stanley Cup futures ticket at +1500 (decimal 16.0). Bankroll at the time was $10,000. The team made the conference final and is now +250 (decimal 3.50) to win the cup. Bankroll today is $11,000 (some other wins along the way).

The half-Kelly framework says don't hedge here. The current edge is so large that even half Kelly puts you above your current exposure level. This is the case for almost all positive-edge futures hedges: the math says don't hedge unless you're at full Kelly to begin with.

If you want to hedge anyway because you want the cash now, that's a behavioral decision, not an EV-optimal one.

The behavioral case for hedging

Pure mathematical optimization says don't hedge. But humans are not purely mathematical optimizers, and hedging often makes sense for non-EV reasons that are still valid:

Sleep value. Carrying a $3,000 swing on a single sports event for two weeks creates real psychological cost. Hedging to flatten the variance can be worth giving up some EV if it lets you function normally during the wait.

Bankroll fragility. A bettor whose betting bankroll is genuinely separate from their living expenses can absorb the variance. A bettor whose betting bankroll is partially their actual savings cannot. Hedging in the second case isn't EV-optimal but it's life-optimal.

One-off windfalls. If you hit a longshot futures ticket that represents months of your normal stake, locking in some realized profit can prevent the situation where the locked-in number would change your life and the unlock-in number wouldn't. The marginal utility of additional dollars at that scale isn't linear, so the strict EV calculation doesn't capture the actual decision.

Capital reallocation. If you have a great new opportunity and your existing futures position is tying up capital, hedging frees up capital for the new bet. Whether this is +EV depends on the new bet's edge vs the EV cost of the hedge vig.

The behavioral case for hedging is real. The math just shouldn't pretend to be the math when it isn't. Hedging at vigged prices is always −EV in the strict sense, and the hedger should understand they're paying for variance reduction. That's a fine trade if you understand what you're trading.

Multi-leg hedging

Sometimes you have multiple bets that combine to create a single complex position, and you want to hedge against a specific outcome rather than against any one bet's failure.

Example. You have a same-game parlay open with three legs (Player A over points, Team B win, Total over). Two legs have hit at halftime. The third leg is still live. You can hedge the third leg by betting the opposite at current live prices, locking in a known profit on the SGP regardless of how the third leg resolves.

Multi-leg hedging is mechanically the same as single-bet hedging, just with the trigger event being "remaining leg(s) hit" rather than "original bet wins." The math doesn't change, only the labeling.

Practical guidance

• Don't hedge to lock in small profit on bets that haven't moved much. The vig you pay isn't worth the variance reduction at small scales.
• Do hedge when a long-running futures bet has appreciated dramatically (5x+) and you have non-betting reasons to want some realized cash.
• Use partial hedging, not full hedging, in most cases. Lock in half of the upside, keep half of the variance. This typically gives you most of the psychological benefit at half the EV cost.
• Compute the hedge price's vig before hedging. If the hedge market has 5% vig, your hedge is paying 2.5% on the hedge stake to flatten variance. Make sure that cost is worth it.
• Track hedged vs unhedged outcomes in your records. Over time you'll see whether your hedging decisions added value (less stress, more reallocated capital) or just cost you EV.
• For most bettors, the right hedging strategy is closer to "rarely hedge" than "always hedge." The defaults that come from books and apps tend to push toward hedging because hedging produces book volume, not because hedging is in the bettor's interest.

The bottom line

Hedging is a tool, not a strategy. The decision to hedge should be driven by what you're actually trying to accomplish: locking in profit, reducing variance, freeing up capital, or managing psychological exposure. Each goal has its own math, and conflating them is the most common mistake in hedging discussions.

EV-optimal hedging is almost never a hedge. Variance-optimal hedging is a partial hedge whose size depends on your tolerance for variance per dollar of EV. Kelly-optimal hedging only applies when your current position size is wrong for the current edge. Behavioral hedging is the most common reason real bettors hedge, and that's fine as long as you understand you're paying for psychological utility, not maximizing dollars.