Actually impact futures
Actually Impact Futures
Impact is how much the market expects an event to change an asset’s price.
It is the market-implied delta between the asset’s price if the event occurs and the asset’s price if it does not occur.
Standard conditional futures do not isolate impact. A long YES and short NO position becomes a bet on which outcome occurs, because one leg settles to the spot price and the other leg settles to 0. Impact exposure instead targets the conditional spread price_yes - price_no. Unconditional outcome futures isolate that spread by making both legs settle to prices in both outcomes, so long YES and short NO settles to impact_TWAP regardless of which outcome occurs. The multiverse or universe token framing is standard conditional futures with different labels.
Definitions
Let an event have two outcomes, YES and NO. Let price_yes(t) be the asset price conditional on YES, and let price_no(t) be the asset price conditional on NO. Define impact(t) as price_yes(t) - price_no(t).
Impact exposure means profit and loss (PnL) depends on the value of impact, not on the event outcome and not on the event probability. A trader can go long impact or short impact. Long impact is a position whose payoff increases when impact increases. Short impact is a position whose payoff increases when impact decreases.
The core problem with “two conditional legs”
In standard conditional token designs, one outcome leg becomes worthless at settlement. That property makes a “long one outcome, short the other outcome” position outcome-dependent. Once the payoff depends on which outcome occurs, the position is exposed to event probability in pricing and risk and one half of the position becomes worthless at resolution.
Unconditional outcome futures: impact exposure using two markets
This mechanism replaces “one leg becomes worthless” with a settlement rule where both outcome markets always settle to a price. It uses S = spot_at_resolution, which is the spot asset price at resolution, and it uses I = impact_TWAP, which is a pre-resolution time-weighted average of impact(t) = price_yes(t) - price_no(t) over an eligibility window.
If YES occurs, the YES leg settles to S, and the NO leg settles to S - I. If NO occurs, the NO leg settles to S, and the YES leg settles to S + I. This makes both legs settle in both outcomes, and it pins the YES-NO spread to I.
Why long YES and short NO pays impact
Consider the position “long one YES unit and short one NO unit”. If YES occurs, the payoff is settle_yes - settle_no = S - (S - I) = I. If NO occurs, the payoff is settle_yes - settle_no = (S + I) - S = I. So “long YES, short NO” pays I regardless of which outcome occurs. That is impact exposure by construction.
Reversing the legs gives short impact. “Short YES, long NO” pays -I regardless of which outcome occurs.
Worked examples
Let S = 100,000 and let I = +10,000. If YES occurs, then settle_yes = 100,000 and settle_no = 90,000, so long YES and short NO pays +10,000. If NO occurs, then settle_yes = 110,000 and settle_no = 100,000, so long YES and short NO pays +10,000.
How impact_TWAP is defined
Impact settlement needs a stable estimate of impact(t) = price_yes(t) - price_no(t).
Using conditional prices when the market becomes one-sided does not give a stable estimate. When p_yes(t) is near 0% or 100%, the low-probability conditional market becomes thin. Low liquidity and low depth make its price cheap to move. A settlement rule that reads that price makes impact easy to manipulate and makes the settled spread unreliable.
The design therefore measures impact only during time where both outcomes have non-trivial probability and both conditional markets carry meaningful liquidity.
It defines p_yes(t) as the YES price in the event market and “in-band time” as time where p_yes(t) ∈ [5%, 95%].
It defines an end time t_end as the event resolution time if p_yes(t) never leaves the band before resolution. Otherwise it defines t_end as the last time before resolution that p_yes(t) leaves the band. This ends the measurement window before the one-sided regime.
It computes impact_TWAP as a 24-hour TWAP of impact(t) over 24 hours of cumulative in-band time ending at t_end.
If there is less than 24 hours of cumulative in-band time prior to t_end, it refunds both outcome markets so settlement does not depend on a short, manipulable window.
Why standard conditional futures do not give impact exposure
Standard conditional futures can also be described using multiverse or universe token language. A deposit creates two outcome claims, and at settlement only the claim for the outcome that occurs redeems to the spot asset price, while the other redeems to 0.
A long YES position and a short NO position of equal size therefore does not settle to impact. The winning conditional future settles to S, and the losing conditional future settles to 0. If YES occurs, the position settles to +S. If NO occurs, it settles to -S. The position is a directional bet on the outcome and on the spot level, not a bet on the pre-event spread price_yes - price_no.
Defining impact as price_yes - spot also does not isolate impact. If spot is priced as a probability-weighted average of conditional prices, spot(t) = p_yes(t)·price_yes(t) + (1-p_yes(t))·price_no(t), then price_yes(t) - spot(t) = (1-p_yes(t))·(price_yes(t) - price_no(t)). The multiplier (1-p_yes(t)) collapses as p_yes(t) approaches 1, so this definition collapses toward 0 as one outcome becomes near-certain.
Scratchpad
If you found this interesting, have feedback or are working on something related, let’s chat: twitter (@distbit0) or schedule a 20 min call
Distbit
futonomous mechanism design. cryptoecon. austrian econ. epistemology. liberty.