Skip to content

Commit 416a24a

Browse files
Soli22deSoli22de
authored
Archive arb persistence research
Merge PR #9 after fixing fill sizing, maker quote, and trade-size assumptions. The corrected research verdict is negative at naive scale and only marginal under cherry-picking, so this lands as research/archive material rather than an execution signal.
1 parent 266b388 commit 416a24a

24 files changed

Lines changed: 4699 additions & 0 deletions

‎reports/arb-persistence-2026-05-12.md‎

Lines changed: 31 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,31 @@
1+
# Arb-persistence Analysis (2026-05-12T14:25:31.893181+00:00)
2+
3+
**Filter**: tier=`explicit_other`, edge_after_fee > 5.00%, max gap 35min
4+
**Snapshot range**: 2026-04-28T21:15:00+00:00 -> 2026-05-12T14:20:04.551587+00:00
5+
**Window**: 329.1 hours across 963 unique snapshots
6+
7+
## Pass / kill metrics
8+
9+
- **Distinct edge events**: 4 (with longtail member: 4)
10+
- **Median persistence (min)**: 2235.0 (P25: 466.4, P75: 3360.0)
11+
12+
**Verdict**: BORDERLINE (extend window or pivot to binary tier)
13+
14+
## Top events by peak edge
15+
16+
| neg_risk_market_id | size | peak_edge | persistence (min) | min_liq seen | longtail | start_ts | end_ts |
17+
|---|---:|---:|---:|---:|:---:|---|---|
18+
| `0xb23e25438839...` | 15 | +0.1823 | 3585.0 | $824 | Y | 2026-05-08T06:15:00+00:00 | 2026-05-10T18:00:00+00:00 |
19+
| `0xb23e25438839...` | 15 | +0.1236 | 1785.0 | $824 | Y | 2026-04-30T18:15:00+00:00 | 2026-05-02T00:00:00+00:00 |
20+
| `0xb23e25438839...` | 15 | +0.0914 | 2685.0 | $824 | Y | 2026-05-05T18:15:00+00:00 | 2026-05-07T15:00:00+00:00 |
21+
| `0xb23e25438839...` | 15 | +0.0893 | 26.8 | $496 | Y | 2026-05-12T13:53:18.607759+00:00 | 2026-05-12T14:20:04.551587+00:00 |
22+
23+
## Notes
24+
25+
- An *event* = contiguous run of snapshots where this group's `edge_after_fee` stayed above the threshold. Gap > max_gap_minutes splits a run.
26+
- `persistence_minutes` = end_ts - start_ts. A one-snapshot flash shows 0 min.
27+
- `min_liquidity_seen` = the minimum group-wide min_liquidity across the event (proxy for thinnest leg).
28+
- For thesis decision use `tier=explicit_other --min-edge 0.05`; rerun with `--tier binary --min-edge 0.02` for the pivot view.
29+
30+
---
31+
*Analyzed at 2026-05-12T14:25:31.893181+00:00*

‎reports/binary-refined-2026-05-12.md‎

Lines changed: 48 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,48 @@
1+
# Refined Binary-tier Analysis (2026-05-12T18:03:42.612726+00:00)
2+
3+
**Source**: today's `markets.ndjson` (question text) + 14-day backfill (`groups.ndjson`).
4+
**Threshold**: edge_after_fee > 2.00%
5+
6+
## 1. Sub-tier counts (today's 2-member groups)
7+
8+
- `dvr`: 71 groups
9+
- `yes_no`: 0 groups
10+
- `pseudo`: 21 groups
11+
12+
## 2. Event metrics by sub-tier
13+
14+
| sub_tier | events | distinct groups | median persistence (min) | top peak edge |
15+
|---|---:|---:|---:|---:|
16+
| `dvr` | 98 | 71 | 1335 | +0.4172 |
17+
| `yes_no` | 0 | 0 | — | — |
18+
| `pseudo` | 27 | 21 | 885 | +0.5192 |
19+
20+
## 3. Top 15 `dvr` events (real D vs R races)
21+
22+
| group | peak_edge | persistence (min) | min_liq | start | end |
23+
|---|---:|---:|---:|---|---|
24+
| `0xc715465f2204...` (Will the Democrats win the Ore vs ...) | +0.4172 | 1785 | $411 | 2026-04-28T21:15:00 | 2026-04-30T03:00:00 |
25+
| `0xa7f79f468a16...` (Will the Democrats win the Sou vs ...) | +0.3849 | 17085 | $367 | 2026-04-28T21:15:00 | 2026-05-10T18:00:00 |
26+
| `0x9bb9ed087667...` (Will the Democrats win the Ark vs ...) | +0.3275 | 885 | $215 | 2026-04-28T21:15:00 | 2026-04-29T12:00:00 |
27+
| `0x266416597e36...` (Will the Republican Party win vs ...) | +0.3251 | 12585 | $206 | 2026-04-28T21:15:00 | 2026-05-07T15:00:00 |
28+
| `0x50a317c8d911...` (Will the Democrats win the Okl vs ...) | +0.2730 | 885 | $122 | 2026-05-03T21:15:00 | 2026-05-04T12:00:00 |
29+
| `0x2aa7cf1991dd...` (Will the Democrats win the Kan vs ...) | +0.2262 | 3585 | $200 | 2026-04-29T12:15:00 | 2026-05-02T00:00:00 |
30+
| `0x5f4893a285ad...` (Will the Democrats win the Wes vs ...) | +0.1979 | 237 | $86 | 2026-05-12T13:53:18 | 2026-05-12T17:50:03 |
31+
| `0x266416597e36...` (Will the Republican Party win vs ...) | +0.1420 | 3585 | $206 | 2026-05-08T06:15:00 | 2026-05-10T18:00:00 |
32+
| `0xf2c1f951aeee...` (Will the Republican Party win vs ...) | +0.1357 | 1785 | $248 | 2026-05-02T15:15:00 | 2026-05-03T21:00:00 |
33+
| `0x9bb9ed087667...` (Will the Democrats win the Ark vs ...) | +0.1296 | 885 | $215 | 2026-04-30T03:15:00 | 2026-04-30T18:00:00 |
34+
| `0xf2c1f951aeee...` (Will the Republican Party win vs ...) | +0.1210 | 4485 | $248 | 2026-05-07T15:15:00 | 2026-05-10T18:00:00 |
35+
| `0xf2c1f951aeee...` (Will the Republican Party win vs ...) | +0.1113 | 3585 | $248 | 2026-05-04T12:15:00 | 2026-05-07T00:00:00 |
36+
| `0x9bb9ed087667...` (Will the Democrats win the Ark vs ...) | +0.1063 | 3585 | $215 | 2026-05-08T06:15:00 | 2026-05-10T18:00:00 |
37+
| `0xb17c29a2fb22...` (Will the Democrats win the Ten vs ...) | +0.1042 | 177 | $2,082 | 2026-05-12T13:53:18 | 2026-05-12T16:50:04 |
38+
| `0x67d0d210eee8...` (Will the Democrats win the Sou vs ...) | +0.0964 | 2685 | $2,176 | 2026-04-28T21:15:00 | 2026-04-30T18:00:00 |
39+
40+
## 4. Notes
41+
42+
- `dvr` (D vs R) is the only sub-tier that is structurally exhaustive in US politics — third parties get < 1% of vote in modern Senate/Governor general elections.
43+
- `yes_no` heuristic is naive (negation-substring + similarity). Likely under-counts.
44+
- `pseudo` events are NOT tradeable as basket arb — the listed 2 members are a sample, not the whole universe.
45+
- Verdict on dvr depends on event count AND median persistence AND a depth check (NOT done yet — same trap as James Bond, where mid edge > bestAsk edge after slippage).
46+
47+
---
48+
*Generated at 2026-05-12T18:03:42.612726+00:00*

‎reports/binary-refined-2026-05-13.md‎

Lines changed: 40 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,40 @@
1+
# Refined Binary-tier Analysis (2026-05-13T03:15:56.289910+00:00)
2+
3+
**Source**: today's `markets.ndjson` (question text) + 14-day backfill (`groups.ndjson`).
4+
**Threshold**: edge_after_fee > 2.00%
5+
6+
## 1. Sub-tier counts (today's 2-member groups)
7+
8+
- `dvr`: 71 groups
9+
- `yes_no`: 0 groups
10+
- `pseudo`: 21 groups
11+
12+
## 2. Event metrics by sub-tier
13+
14+
| sub_tier | events | distinct groups | median persistence (min) | top peak edge |
15+
|---|---:|---:|---:|---:|
16+
| `dvr` | 7 | 71 | 15 | +0.1979 |
17+
| `yes_no` | 0 | 0 | — | — |
18+
| `pseudo` | 7 | 20 | 570 | +0.0704 |
19+
20+
## 3. Top 15 `dvr` events (real D vs R races)
21+
22+
| group | peak_edge | persistence (min) | min_liq | start | end |
23+
|---|---:|---:|---:|---|---|
24+
| `0x5f4893a285ad...` (Will the Democrats win the Wes vs ...) | +0.1979 | 792 | $86 | 2026-05-12T13:53:18 | 2026-05-13T03:05:04 |
25+
| `0xb17c29a2fb22...` (Will the Democrats win the Ten vs ...) | +0.1042 | 177 | $2,082 | 2026-05-12T13:53:18 | 2026-05-12T16:50:04 |
26+
| `0xa8574c0caacc...` (Will the Democrats win the Sou vs ...) | +0.0369 | 765 | $2,125 | 2026-05-12T14:20:04 | 2026-05-13T03:05:04 |
27+
| `0x50a317c8d911...` (Will the Democrats win the Okl vs ...) | +0.0355 | 15 | $187 | 2026-05-12T15:35:04 | 2026-05-12T15:50:03 |
28+
| `0x50a317c8d911...` (Will the Democrats win the Okl vs ...) | +0.0335 | 15 | $143 | 2026-05-12T16:20:04 | 2026-05-12T16:35:04 |
29+
| `0x50a317c8d911...` (Will the Democrats win the Okl vs ...) | +0.0310 | 0 | $1,205 | 2026-05-12T18:50:03 | 2026-05-12T18:50:03 |
30+
| `0x50a317c8d911...` (Will the Democrats win the Okl vs ...) | +0.0259 | 0 | $192 | 2026-05-12T18:05:03 | 2026-05-12T18:05:03 |
31+
32+
## 4. Notes
33+
34+
- `dvr` (D vs R) is the only sub-tier that is structurally exhaustive in US politics — third parties get < 1% of vote in modern Senate/Governor general elections.
35+
- `yes_no` heuristic is naive (negation-substring + similarity). Likely under-counts.
36+
- `pseudo` events are NOT tradeable as basket arb — the listed 2 members are a sample, not the whole universe.
37+
- Verdict on dvr depends on event count AND median persistence AND a depth check (NOT done yet — same trap as James Bond, where mid edge > bestAsk edge after slippage).
38+
39+
---
40+
*Generated at 2026-05-13T03:15:56.289910+00:00*

‎reports/experiment-multi-model-results-glm-4-6-2026-05-12.md‎

Lines changed: 29 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,29 @@
1+
# T2 多模型对比实验报告（2026-05-12T08:48:17.704098+00:00）
2+
3+
**Sample**: n=30 markets stratified by description length
4+
**Models**: 1 via OpenRouter
5+
**Total cost**: $0.0000
6+
7+
## 1. 自动指标对比
8+
9+
| 模型 | 调用 | schema_ok | grounding_ok | nonempty | clauses/市场 | 平均 in/out tok | 平均延迟 | 总成本 | $/call |
10+
|---|---:|---:|---:|---:|---:|---:|---:|---:|---:|
11+
| GLM-4.6 (elysiver) | 30 | 29/29 | 29/29 | 29/29 | 3.3 | 537/409 | 13.0s | $0.0000 | $0.00000 |
12+
13+
## 2. 按 description 长度桶分布（clauses/市场）
14+
15+
| 模型 | short (100-300) | medium (300-700) | long (700+) |
16+
|---|---:|---:|---:|
17+
| GLM-4.6 (elysiver) | 2.1 | 3.6 | 4.2 |
18+
19+
## 3. 错误清单
20+
21+
- `glm-4.6` on `579379`: URL error: [WinError 10060] 由于连接方在一段时间后没有正确答复或连接的主机没有反应，连接尝试失败。
22+
23+
## 4. 数据归档
24+
25+
完整 per-call 结果在 `data\experiments\2026-05-12\multi-model-results-glm-4-6.ndjson` (30 行)。
26+
下一步：人工 + Claude 裁判模式，按 actionable / structural / trivial 评分各模型 clauses。
27+
28+
---
29+
*Snapshot: 2026-05-12T08:48:17.704098+00:00*

‎reports/experiment-multi-model-results-glm-5-2026-05-12.md‎

Lines changed: 42 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,42 @@
1+
# T2 多模型对比实验报告（2026-05-12T08:55:14.783393+00:00）
2+
3+
**Sample**: n=30 markets stratified by description length
4+
**Models**: 1 via OpenRouter
5+
**Total cost**: $0.0000
6+
7+
## 1. 自动指标对比
8+
9+
| 模型 | 调用 | schema_ok | grounding_ok | nonempty | clauses/市场 | 平均 in/out tok | 平均延迟 | 总成本 | $/call |
10+
|---|---:|---:|---:|---:|---:|---:|---:|---:|---:|
11+
| GLM-5 (elysiver) | 30 | 16/16 | 16/16 | 16/16 | 3.9 | 448/402 | 40.9s | $0.0000 | $0.00000 |
12+
13+
## 2. 按 description 长度桶分布（clauses/市场）
14+
15+
| 模型 | short (100-300) | medium (300-700) | long (700+) |
16+
|---|---:|---:|---:|
17+
| GLM-5 (elysiver) | 3.0 | 4.2 | 4.6 |
18+
19+
## 3. 错误清单
20+
21+
- `glm-5` on `631187`: HTTP 504: error code: 504
22+
- `glm-5` on `676843`: HTTP 504: error code: 504
23+
- `glm-5` on `676745`: HTTP 504: error code: 504
24+
- `glm-5` on `676741`: HTTP 504: error code: 504
25+
- `glm-5` on `679504`: HTTP 504: error code: 504
26+
- `glm-5` on `687272`: HTTP 504: error code: 504
27+
- `glm-5` on `620682`: HTTP 504: error code: 504
28+
- `glm-5` on `678930`: HTTP 504: error code: 504
29+
- `glm-5` on `558975`: HTTP 504: error code: 504
30+
- `glm-5` on `631125`: HTTP 504: error code: 504
31+
- `glm-5` on `701757`: HTTP 504: error code: 504
32+
- `glm-5` on `679028`: HTTP 504: error code: 504
33+
- `glm-5` on `577177`: HTTP 504: error code: 504
34+
- `glm-5` on `618508`: HTTP 504: error code: 504
35+
36+
## 4. 数据归档
37+
38+
完整 per-call 结果在 `data\experiments\2026-05-12\multi-model-results-glm-5.ndjson` (30 行)。
39+
下一步：人工 + Claude 裁判模式，按 actionable / structural / trivial 评分各模型 clauses。
40+
41+
---
42+
*Snapshot: 2026-05-12T08:55:14.783393+00:00*

‎reports/experiment-negrisk-mispricing-2026-05-12.md‎

Lines changed: 110 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,110 @@
1+
# 长尾 Neg-Risk Mispricing 普查实验 7 (refined)（2026-05-12T10:19:18.834485+00:00）
2+
3+
**v2 改进**：原版 `likely_exhaustive` 把 size≥8 当作 exhaustive 是错的（Nobel 奖那种 20 人候选其实根本不穷举）。
4+
本版分级：
5+
- `explicit_other` —— 组内含 'No one / None / Another' 等显式 catch-all member（高置信度真 exhaustive）
6+
- `binary` —— 正好 2 个 member（多数是 D/R 政治对决，**可能**穷举，但有第三方风险）
7+
- `open_set` —— 3+ member 且无 catch-all（**几乎确定不穷举**，basket arb 是假信号）
8+
9+
---
10+
11+
## 1. 重新分级后的基本计数
12+
13+
- 总 neg-risk 组（已过滤 ask 退化）：**151**
14+
- `explicit_other`（**高置信度 exhaustive**）：**1** (1%)
15+
- `binary` (2 member)：**73** (48%)
16+
- `open_set` (假阳性源)：**77** (51%)
17+
- 含至少 1 个长尾成员（vol24hr < $40）：**99** (66%)
18+
19+
## 2. Sum(YES_ask) < 1.0 + edge>0 三分类
20+
21+
- **Strict 候选**（confidently exhaustive + edge_after_fee > 0）：**1**
22+
- **Binary 候选**（2 member + edge > 0，需验证第三方风险）：**7**
23+
- **Open-set 假阳性**（看似 edge > 0 但 basket 不穷举，**不能交易**）：**6**
24+
25+
## 3. Strict 候选（真正值得 follow-up）
26+
27+
| Group | size | sum_ask | fee_total | edge_after_fee | longtail? | tier | min_liq |
28+
|---|---:|---:|---:|---:|:---:|:---:|---:|
29+
| `0xb23e25438839…` | 15 | 0.8930 | 0.01772 | +0.0893 | ✓ | explicit | $2,210 |
30+
31+
### Strict #1: edge = +0.0893, longtail = True
32+
33+
| Member | bestAsk | bestBid | vol24hr | liquidity | fee_rate |
34+
|---|---:|---:|---:|---:|---:|
35+
| Aaron Taylor-Johnson announced as next James Bond? | 0.016 | 0.005 | $113 | $2,210 | 0.0500 |
36+
| James Norton announced as next James Bond? | 0.004 | 0.002 | $512 | $3,210 | 0.0500 |
37+
| Paul Mescal announced as next James Bond? | 0.002 | 0.001 | $30 | $4,803 | 0.0500 |
38+
| Jacob Elordi announced as next James Bond? | 0.037 | 0.035 | $72 | $3,349 | 0.0500 |
39+
| Harris Dickinson announced as next James Bond? | 0.011 | 0.002 | $289 | $3,985 | 0.0500 |
40+
| Tom Hardy announced as next James Bond? | 0.009 | 0.001 | $30 | $2,582 | 0.0500 |
41+
| Pierce Brosnan announced as next James Bond? | 0.002 | 0.001 | $113 | $5,743 | 0.0500 |
42+
| Tom Holland announced as next James Bond? | 0.012 | 0.007 | $30 | $3,173 | 0.0500 |
43+
| Henry Cavill announced as next James Bond? | 0.002 | 0.001 | $62 | $5,696 | 0.0500 |
44+
| Callum Turner announced as next James Bond? | 0.060 | 0.032 | $746 | $3,526 | 0.0500 |
45+
| Jack Lowdon announced as next James Bond? | 0.002 | 0.001 | $30 | $5,933 | 0.0500 |
46+
| Theo James announced as next James Bond? | 0.002 | 0.001 | $530 | $3,336 | 0.0500 |
47+
| James Collier announced as next James Bond? | 0.002 | 0.001 | $114 | $5,332 | 0.0500 |
48+
| Josh O'Connor announced as next James Bond? | 0.002 | 0.001 | $30 | $3,390 | 0.0500 |
49+
| No one announced as next James Bond? 🅾️ | 0.730 | 0.7 | $62 | $2,527 | 0.0500 |
50+
51+
## 4. Binary 候选（edge > 0，第三方风险待评估）
52+
53+
| Group | sum_ask | edge_after_fee | longtail? | min_liq |
54+
|---|---:|---:|:---:|---:|
55+
| `0x5f4893a285ad…` | 0.9640 | +0.0314 | ✓ | $2,634 |
56+
| `0xb17c29a2fb22…` | 0.9660 | +0.0275 | | $2,722 |
57+
| `0xafccb4ac9586…` | 0.9890 | +0.0093 | ✓ | $13,622 |
58+
| `0xe0ff15139f33…` | 0.9880 | +0.0079 | ✓ | $10,987 |
59+
| `0xa8574c0caacc…` | 0.9900 | +0.0051 | ✓ | $4,009 |
60+
| `0x9bb9ed087667…` | 0.9930 | +0.0035 | ✓ | $2,527 |
61+
| `0x67d0d210eee8…` | 0.9940 | +0.0010 | ✓ | $3,375 |
62+
63+
Binary 组样本（前 3 个完整 member 表）：
64+
65+
#### Binary #1: edge = +0.0314
66+
| Member | bestAsk | bestBid | vol24hr | liquidity | fee_rate |
67+
|---|---:|---:|---:|---:|---:|
68+
| Will the Democrats win the West Virginia Senate race in 2026 | 0.044 | 0.029 | $0 | $2,634 | 0.0400 |
69+
| Will the Republicans win the West Virginia Senate race in 20 | 0.920 | 0.9 | $5 | $3,864 | 0.0400 |
70+
71+
#### Binary #2: edge = +0.0275
72+
| Member | bestAsk | bestBid | vol24hr | liquidity | fee_rate |
73+
|---|---:|---:|---:|---:|---:|
74+
| Will the Democrats win the Tennessee governor race in 2026? | 0.072 | 0.051 | $110 | $6,776 | 0.0400 |
75+
| Will the Republicans win the Tennessee governor race in 2026 | 0.894 | 0.89 | $100 | $2,722 | 0.0400 |
76+
77+
#### Binary #3: edge = +0.0093
78+
| Member | bestAsk | bestBid | vol24hr | liquidity | fee_rate |
79+
|---|---:|---:|---:|---:|---:|
80+
| Will Kathy Hochul win the 2026 New York Democratic Gubernato | 0.973 | 0.971 | $71 | $16,225 | 0.0400 |
81+
| Will Antonio Delgado win the 2026 New York Democratic Gubern | 0.016 | 0.008 | $0 | $13,622 | 0.0400 |
82+
83+
## 5. Open-set 假阳性（**不是机会**，列出避免误导）
84+
85+
这些组 sum < 1 看似有 arb edge，但成员列表**不穷举** —— 实际胜者不在列表里时整篮归零。**不要根据这些数据交易。**
86+
87+
| Group | size | sum_ask | edge_after_fee | sample question |
88+
|---|---:|---:|---:|---|
89+
| `0x09139fb03e82…` | 20 | 0.4590 | +0.5191 | Will Donald Trump win the Nobel Peace Prize in 202… |
90+
| `0x03c9c6d9f7b2…` | 17 | 0.9380 | +0.0620 | Will Benjamin Netanyahu be the next Prime Minister… |
91+
| `0x92fa5dc99566…` | 12 | 0.9610 | +0.0390 | Will Israel strike 4 countries in 2026?… |
92+
| `0x4d76347d4d1e…` | 7 | 0.9810 | +0.0073 | Will NVIDIA be the largest company in the world by… |
93+
| `0x80ee92187ee5…` | 14 | 0.9930 | +0.0026 | Will Kim Dong-yeon win the 2026 Gyeonggi Province … |
94+
| `0x0f630f2d9401…` | 3 | 0.9800 | +0.0011 | Will the Democrats win the Michigan governor race … |
95+
96+
## 6. 重要警告（读结果前必看）
97+
98+
- **bestAsk 是 snapshot 时刻的最优挂单价**，不是任意可成交数量的均价。Long-tail 市场上 bestAsk 后面可能只挂 $5 的深度。
99+
- **fee_rate** 取每市场自带的 `feeSchedule.rate`；feesEnabled=False 的市场记 0。
100+
- **strict 也只是 'confidently exhaustive' 的启发式判定**，不是结构保证。生产时需要看 Polymarket Neg Risk Adapter 合约的实际配置。
101+
- **slippage 可能吃掉所有 edge** —— 实际交易前需要用 CLOB `/book` 端点核对深度。
102+
103+
## 7. 真正要回答的问题
104+
105+
- 今天 snapshot 下，**confidently exhaustive 且 edge > 0 after fees 的组数 = 1**。
106+
- 长尾子集 = 1
107+
- 进入下一步的条件：用 CLOB orderbook 复核这 1 个候选的真实可成交深度，剔除 slippage 吃掉 edge 的，剩多少是真 alpha。
108+
109+
---
110+
*Snapshot: 2026-05-12T10:19:18.834485+00:00*

‎reports/group-book-validation-will-the-democrats-win-the-2026-05-13.md‎

Lines changed: 31 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,31 @@
1+
# Group Book Validation: will-the-democrats-win-the (2026-05-13T06:27:27.152124+00:00)
2+
3+
Real-orderbook depth check of basket arb candidate.
4+
Group: `0xa8574c0caacc...` (members=2)
5+
6+
## Per-member top-of-book
7+
8+
| # | Question | bestAsk | depth @ bestAsk | bestBid | gamma_ask | fee | vol24hr | liq |
9+
|---|---|---:|---:|---:|---:|---:|---:|---:|
10+
| 1 | Will the Democrats win the South Carolina governor race | 0.059 | 77 | 0.022 | 0.0590 | 0.040 | $0 | $4,664 |
11+
| 2 | Will the Republicans win the South Carolina governor ra | 0.91 | 4 | 0.89 | 0.9100 | 0.040 | $1 | $5,585 |
12+
13+
## Marginal (1-unit) edge
14+
15+
- bestAsk basket sum: **0.9690**
16+
- marginal fee: 0.00550
17+
- marginal edge_after_fee: **+0.0255**
18+
19+
## Fill simulation (fix: cost & edge computed at ACTUAL fillable size, not intended)
20+
21+
| Intended size | Actual fillable | Avg cost/unit | Total fee | Edge $ | Edge % |
22+
|---:|---:|---:|---:|---:|---:|
23+
| 50 | 50.0 | 0.9782 | $0.26 | $+0.83 | +1.66% |
24+
| 200 | 200.0 | 0.9845 | $1.07 | $+2.02 | +1.01% |
25+
| 500 | 500.0 | 1.0178 | $3.24 | $-12.12 | -2.42% |
26+
| 1000 | 1000.0 | 1.0430 | $6.53 | $-49.48 | -4.95% |
27+
| 2000 | 1303.9 ⚠️ | 1.0566 | $8.26 | $-82.06 | -6.29% |
28+
| 5000 | 1303.9 ⚠️ | 1.0566 | $8.26 | $-82.06 | -6.29% |
29+
30+
---
31+
*Snapshot: 2026-05-13T06:27:27.152124+00:00*

0 commit comments

Comments
 (0)