秋田预测市场 · Discord 审计

250ms taker 延迟引发交易者反弹与延迟优化讨论

2026-07-17 · Polymarket · 55 条相关讨论

Coolnyul 2026-07-16 20:01:39

yes, with a p90 cancel time of a few hundred ms

yes, with a p90 cancel time of a few hundred ms

The Dude 2026-07-16 20:01:41

Yes but cancels take 500ms dw

Yes but cancels take 500ms dw

James 2026-07-16 20:13:10

LOL, so 250 ms is still nothing.

LOL, so 250 ms is still nothing.

WildRage 2026-07-16 20:23:58

How does anyone even make if that's the case?

How does anyone even make if that's the case?

tehnoobkiddo 2026-07-16 20:26:21

everybody gets delayed

everybody gets delayed

TeenyTinySharpe 2026-07-16 20:33:22

taker delay of 250ms is also a minimum

taker delay of 250ms is also a minimum

TeenyTinySharpe 2026-07-16 20:33:30

reality is taker latencies often go p50 1s+

reality is taker latencies often go p50 1s+

richlb 2026-07-16 21:11:23

We don't need taker delay

We don't need taker delay

richlb 2026-07-16 21:11:28

let's remove it

let's remove it

richlb 2026-07-16 21:12:00

😈 kalshi have TWAP so polymarket must have TWAP then kalshi don't have taker delay, why polymarket should have

😈 kalshi have TWAP so polymarket must have TWAP then kalshi don't have taker delay, why polymarket should have

richlb 2026-07-16 21:13:31

I'm going to let Kalshi guys to mention "taker delay is bad, polymarket have , we don't" more on Twitter, then Polymarket might drop taker delay.

I'm going to let Kalshi guys to mention "taker delay is bad, polymarket have , we don't" more on Twitter, then Polymarket might drop taker delay.

The Dude 2026-07-16 21:27:26

Once polymarket is exactly the same as Kalshi but infra is worse, no reason for ppl to stay here

Once polymarket is exactly the same as Kalshi but infra is worse, no reason for ppl to stay here

Aemond Targaryen 2026-07-16 21:31:18

I reckon there's more "smart dumb" money on polymarket still

I reckon there's more "smart dumb" money on polymarket still

weathergirl 2026-07-16 22:28:11

how are you avoiding adverse fills?

how are you avoiding adverse fills?

Envoyoflight 2026-07-16 22:42:17

mate 250 is what they say it is. in reality it's worse

mate 250 is what they say it is. in reality it's worse

weathergirl 2026-07-17 00:42:42

ok ok ok - continue taking my money 🙇‍♂️

ok ok ok - continue taking my money 🙇‍♂️

weathergirl 2026-07-17 00:43:17

if you already wanna flex here about not having issues with it, you should at least help us a tiny bit

if you already wanna flex here about not having issues with it, you should at least help us a tiny bit

The Dude 2026-07-17 00:58:17

Well you have high fees on PM like 2-4% plus the taker delay plus 5m markets are quite efficient

Well you have high fees on PM like 2-4% plus the taker delay plus 5m markets are quite efficient

Chien mechant 2026-07-17 01:26:56

It's a speed game you are not the fastest plus taker speed for available price is like 1500 ms

It's a speed game you are not the fastest plus taker speed for available price is like 1500 ms

James 2026-07-17 01:44:59

That’s a good explanation. But the 250 ms maker protection has been in place for months, and this strategy has only become unfeasible since last month.

That’s a good explanation. But the 250 ms maker protection has been in place for months, and this strategy has only become unfeasible since last month.

ELOSANTA 2026-07-17 01:45:23

its only unfeasible since you pay high fees + polymarket nerfed referral rewards

its only unfeasible since you pay high fees + polymarket nerfed referral rewards

ELOSANTA 2026-07-17 01:45:46

so the bigger winners win more

so the bigger winners win more

ELOSANTA 2026-07-17 01:45:51

and if u pay high fees u cant enter it

and if u pay high fees u cant enter it

ELOSANTA 2026-07-17 01:45:57

kind of a monopoly

kind of a monopoly

James 2026-07-17 01:54:35

I’ve optimized the server location and the code path for preparing signatures, and I submit same order N time asynchronously. The only thing I can imagine being faster than me is an internal API endpoint.

I’ve optimized the server location and the code path for preparing signatures, and I submit same order N time asynchronously. The only thing I can imagine being faster than me is an internal API endpoint.

James 2026-07-17 02:16:19

The fastest connection might save us 20 ms, but it still can’t overcome the 250 ms maker-protection window. They can always cancel their orders.

The fastest connection might save us 20 ms, but it still can’t overcome the 250 ms maker-protection window. They can always cancel their orders.

Cayden 2026-07-17 02:32:30

What is the best proxy/aws server to have the lowest latency

What is the best proxy/aws server to have the lowest latency

The Dude 2026-07-17 02:47:13

where the CLOB is?

where the CLOB is?

Cayden 2026-07-17 02:47:51

obviously I don't know where the CLOB is located so that's why I'm asking

obviously I don't know where the CLOB is located so that's why I'm asking

The Dude 2026-07-17 02:50:49

or you can just spin up one us-east-1 and one eu-west-2 instance and ping the server

or you can just spin up one us-east-1 and one eu-west-2 instance and ping the server

javi 2026-07-17 03:29:15

Im getting ghost fills again

Im getting ghost fills again

javi 2026-07-17 03:29:19

anyone else?

anyone else?

Kawa 2026-07-17 08:01:37

<@1361217824776786031> twap gon cut this to 10k real quick btw

<@1361217824776786031> twap gon cut this to 10k real quick btw

Kawa 2026-07-17 08:02:05

and its not gon be cuttin out the manipulators

and its not gon be cuttin out the manipulators

Kawa 2026-07-17 08:02:11

lmao acc this might be genius

lmao acc this might be genius

Kawa 2026-07-17 08:02:23

kill volume, disincentivize manipulators from manipulating

kill volume, disincentivize manipulators from manipulating

Kawa 2026-07-17 08:02:56

idk whats to learn

idk whats to learn

Kawa 2026-07-17 08:03:01

we been yellin at them for months

we been yellin at them for months

Kawa 2026-07-17 08:03:05

they clearly dgaf

they clearly dgaf

gmdavid120 2026-07-17 08:03:11

manip wont care bout twap

manip wont care bout twap

Kawa 2026-07-17 08:03:13

if their best solution to this

if their best solution to this

Kawa 2026-07-17 08:03:20

their BEST solution is twap

their BEST solution is twap

gmdavid120 2026-07-17 08:03:21

so tehy wont be killed

so tehy wont be killed

richlb 2026-07-17 08:03:28

About TWAP, I have a idea. <:pepecry:1480174599093096560> I think Polymarket only needs to adopt the same TWAP rules as Kalshi on 15m markets, while keeping 5m on the current rules for now. The upside: 1. Once 15m rules align with Kalshi, liquidity should start sharing immediately. 2. 5m stays as-is, so users still get choice: one product that matches the competitor, one that keeps drawing thrill-seekers. Right now 5m and 15m on Polymarket overlap completely, and 15m has almost no liquidity. If both switch to TWAP the same way, 5m will still cannibalize 15m liquidity.

About TWAP, I have a idea. <:pepecry:1480174599093096560> I think Polymarket only needs to adopt the same TWAP rules as Kalshi on 15m markets, while keeping 5m on the current rules for now. The upside: 1. Once 15m rules align with Kalshi, liquidity should start sharing immediately. 2. 5m stays as-is, so users still get choice: one product that matches the competitor, one that keeps drawing thrill-seekers. Right now 5m and 15m on Polymarket overlap completely, and 15m has almost no liquidity. If both switch to TWAP the same way, 5m will still cannibalize 15m liquidity.

Kawa 2026-07-17 08:04:02

thats based on the assumption that ppl gaf about twap

thats based on the assumption that ppl gaf about twap

Kawa 2026-07-17 08:04:11

ur assuming a mass migration will happen to twap

ur assuming a mass migration will happen to twap

Kawa 2026-07-17 08:04:28

thats not why kalshi has its volume

thats not why kalshi has its volume

Kawa 2026-07-17 08:04:51

they have volume cuz of user trust which poly no longer has for crypto

they have volume cuz of user trust which poly no longer has for crypto

Kawa 2026-07-17 08:11:06

<@1361217824776786031> bring back 500ms taker delay

<@1361217824776786031> bring back 500ms taker delay

Kawa 2026-07-17 08:11:22

either bring back the delay or lower the fees

either bring back the delay or lower the fees

Kawa 2026-07-17 08:11:25

one or the other

one or the other

Nemo 2026-07-17 08:13:19

Theyre great

Theyre great

Nemo 2026-07-17 08:13:24

But dead true

But dead true