秋田预测市场 · Discord 审计

060610的做市策略实盘逆向选择分析

2026-07-04 · Polymarket · 21 条相关讨论

Hundt 2026-07-04 16:12:17

new strat still fighting. <@650765325645905940> pls tell me what you think. Also say its just luck if it looks like it

new strat still fighting. <@650765325645905940> pls tell me what you think. Also say its just luck if it looks like it

033033033 2026-07-04 16:13:02

Interesting graph

Interesting graph

033033033 2026-07-04 16:13:16

Im not really seeing any kind of trend

Im not really seeing any kind of trend

Hundt 2026-07-04 16:13:42

No Im not either.

No Im not either.

Hundt 2026-07-04 16:13:49

Wait this makes it more clear:

Wait this makes it more clear:

Hundt 2026-07-04 16:14:20

this is by time

this is by time

Hundt 2026-07-04 16:14:47

It mostlytrades in volatile phases and bleeds in calm marketphases where exchanges are closed.

It mostlytrades in volatile phases and bleeds in calm marketphases where exchanges are closed.

Hundt 2026-07-04 16:15:00

but also only 5 daysamplesize

but also only 5 daysamplesize

X 2026-07-04 16:16:00

use some sort of volatile filter and then backtest it again

use some sort of volatile filter and then backtest it again

Hundt 2026-07-04 16:16:33

but maybe my volatile filter is just not that good

but maybe my volatile filter is just not that good

Hundt 2026-07-04 16:24:39

look the adverse selection in comparison to the paper trader. They really move in parallel most of thetime. But there are specific times, where live gets adverse selectet a lot. Maybe thats fixable

look the adverse selection in comparison to the paper trader. They really move in parallel most of thetime. But there are specific times, where live gets adverse selectet a lot. Maybe thats fixable

Penps 2026-07-04 16:27:57

But it also seems like spikes come after you get adverse selected

But it also seems like spikes come after you get adverse selected

Hundt 2026-07-04 16:28:45

I am running the study atm. I need a bigger samplesize before I can decide if its noise or a real signal

I am running the study atm. I need a bigger samplesize before I can decide if its noise or a real signal

Penps 2026-07-04 16:29:22

I’m currently at the same point at you are

I’m currently at the same point at you are

Penps 2026-07-04 16:30:05

I’m checking either volatility regimes or time based rn waiting for results

I’m checking either volatility regimes or time based rn waiting for results

Coolnyul 2026-07-04 17:06:13

Impressive how the shape is the same in live as in backtest. Is it mostly taker?

Impressive how the shape is the same in live as in backtest. Is it mostly taker?

Hundt 2026-07-04 17:31:02

Only maker. And Backtest looks better.

Only maker. And Backtest looks better.

Coolnyul 2026-07-04 17:37:49

Yeah, but it is really difficult to make a backtest the same shape as the real for a maker strategy

Yeah, but it is really difficult to make a backtest the same shape as the real for a maker strategy

Ban 2026-07-04 17:38:57

you full maker?

you full maker?

Ban 2026-07-04 17:39:39

how do you backtest a maker strat monkaS

how do you backtest a maker strat monkaS

TiCkShot 2026-07-04 18:18:21

You can't as accurately simulate queue mechanics in paper mode, that's what makes it hard in live

You can't as accurately simulate queue mechanics in paper mode, that's what makes it hard in live