arbitrage-trading

Name: arbitrage-trading
Author: SKE-Labs

15stars2forks

How to install arbitrage-trading?

ske-labs/agent-trading-skills/arbitrage-trading

$npx -y skills add SKE-Labs/agent-trading-skills --skill arbitrage-trading

Installs into the current project.

Summary

Screen execution-sensitive price differences across exchanges, pairs, spot, and derivatives. Use when evaluating synchronized spreads, triangular conversion paths, or basis trades after fees, depth, settlement, counterparty, and inventory risk.

Files · 1

View on GitHub

SKILL.md

1	# Arbitrage Trading
2
3	Identify executable price differences while assuming that every apparent spread may compensate for latency, liquidity, settlement, or venue risk.
4
5	> Note: Agent can identify arbitrage opportunities via price comparison but cannot execute cross-exchange trades directly. Use this skill to detect and report opportunities.
6
7	## Arbitrage Types
8
9	\| Type \| Mechanism \| Key Cost Factor \|
10	\| ---------------- \| ------------------------------------------ \| ----------------- \|
11	\| Cross-Exchange \| Same asset priced differently on two venues \| Transfer fees \|
12	\| Triangular \| Three-pair cycle (BTC->ETH->USDT->BTC) \| Trading fees x3 \|
13	\| Futures-Spot \| Premium between spot and perp/futures \| Funding rate \|
14	\| DEX-CEX \| Price gap between decentralized and centralized \| Gas fees \|
15
16	## Profitability Check
17
18	Cross-Exchange:
19	- Gross P&L = executable sell proceeds − executable buy cost for the same quantity
20	- Net P&L = Gross P&L − fees − slippage − funding/borrow − transfer/gas − hedge and rebalancing costs
21
22	Triangular:
23	- Expected cross rate = Price_A / Price_B
24	- Actual cross rate = observed market rate
25	- Spread % = (Actual - Expected) / Expected x 100
26
27	Require a positive buffer above estimated costs and model error; `Net Profit > 0` on stale top-of-book quotes is not sufficient.
28
29	## Workflow
30
31	1. Fetch synchronized executable quotes from every leg. Candle closes are a coarse screening proxy only; do not call the result actionable without bid/ask, depth, size, timestamp, and venue status.
32	```
33	get_candles(symbol="BTC/USD", exchange="binance", interval="5m", count=1)
34	get_candles(symbol="ETH/USD", exchange="binance", interval="5m", count=1)
35	get_candles(symbol="ETH/BTC", exchange="binance", interval="5m", count=1)
36	```
37
38	2. Simulate the exact path using the correct bid or ask at each leg, lot/tick rules, rounding, and available depth.
39
40	3. Check carry and convergence terms from the venue's contract specification, funding history, margin rules, and settlement index—not a news summary alone:
41	```
42	get_financial_news(topic="BTC funding rate perpetual futures premium")
43	```
44
45	4. Stress all costs and failures: one-leg fill, reject, latency, transfer halt, depeg, funding flip, liquidation, borrow recall, and exchange default.
46
47	5. Report opportunity with: pair(s), spread %, estimated fees, net profit, and time sensitivity.
48
49	## Triangular Arbitrage Example
50
51	```
52	BTC/USDT: 50,000 \| ETH/USDT: 2,000 \| ETH/BTC: 0.041
53	Expected ETH/BTC: 2000/50000 = 0.040
54	Actual: 0.041 → ~2.5% spread (before fees)
55	Path: USDT → BTC → ETH → USDT
56	```
57
58	## Evidence and Validation
59
60	- Treat the setup as a testable hypothesis, not a prediction. Define thresholds, entry, invalidation, and exit before evaluating outcomes.
61	- Calibrate on the same instrument, venue, session, and timeframe. Use closed candles and a held-out or walk-forward sample; record every variant tried.
62	- Include spread, fees, slippage, borrow or funding, partial fills, and latency. Reject the setup when net expectancy is not positive or depends on one narrow parameter.
63	- Return observed inputs, missing data, cost assumptions, entry, invalidation, exit, and a valid, watch, or no-trade status.
64	- Research basis: [John, Li & Liu](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4816710) and [Vidal-Tomás](https://www.sciencedirect.com/science/article/pii/S154461232401393X) show that observed crypto spreads reflect liquidity, settlement, fees, and exchange-default risk—not risk-free profit.
65
66	## Key Rules
67
68	- NEVER report an opportunity without deducting all fees (trading, transfer, gas, slippage)
69	- NEVER assume execution is instant -- note that cross-exchange opportunities decay in seconds
70	- Pre-funded accounts on both sides are required for cross-exchange arb; transfers kill the edge
71	- Use a venue-, size-, and latency-specific minimum edge; no universal spread floor is valid
72	- Check order book depth: thin liquidity means slippage will eat the spread
73
74	## Related Skills
75
76	- funding-rate-trading -- funding rate arb (spot + perp) is a specific delta-neutral strategy
77	- altcoin-rotation -- cross-pair price analysis overlaps with rotation screening

People who liked this also viewed

6 picks

Type

arbitrage-trading

15stars2forks

View on GitHub

How to install arbitrage-trading?

ske-labs/agent-trading-skills/arbitrage-trading

$npx -y skills add SKE-Labs/agent-trading-skills --skill arbitrage-trading

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Arbitrage Trading
2
3	Identify executable price differences while assuming that every apparent spread may compensate for latency, liquidity, settlement, or venue risk.
4
5	> Note: Agent can identify arbitrage opportunities via price comparison but cannot execute cross-exchange trades directly. Use this skill to detect and report opportunities.
6
7	## Arbitrage Types
8
9	\| Type \| Mechanism \| Key Cost Factor \|
10	\| ---------------- \| ------------------------------------------ \| ----------------- \|
11	\| Cross-Exchange \| Same asset priced differently on two venues \| Transfer fees \|
12	\| Triangular \| Three-pair cycle (BTC->ETH->USDT->BTC) \| Trading fees x3 \|
13	\| Futures-Spot \| Premium between spot and perp/futures \| Funding rate \|
14	\| DEX-CEX \| Price gap between decentralized and centralized \| Gas fees \|
15
16	## Profitability Check
17
18	Cross-Exchange:
19	- Gross P&L = executable sell proceeds − executable buy cost for the same quantity
20	- Net P&L = Gross P&L − fees − slippage − funding/borrow − transfer/gas − hedge and rebalancing costs
21
22	Triangular:
23	- Expected cross rate = Price_A / Price_B
24	- Actual cross rate = observed market rate
25	- Spread % = (Actual - Expected) / Expected x 100
26
27	Require a positive buffer above estimated costs and model error; `Net Profit > 0` on stale top-of-book quotes is not sufficient.
28
29	## Workflow
30
31	1. Fetch synchronized executable quotes from every leg. Candle closes are a coarse screening proxy only; do not call the result actionable without bid/ask, depth, size, timestamp, and venue status.
32	```
33	get_candles(symbol="BTC/USD", exchange="binance", interval="5m", count=1)
34	get_candles(symbol="ETH/USD", exchange="binance", interval="5m", count=1)
35	get_candles(symbol="ETH/BTC", exchange="binance", interval="5m", count=1)
36	```
37
38	2. Simulate the exact path using the correct bid or ask at each leg, lot/tick rules, rounding, and available depth.
39
40	3. Check carry and convergence terms from the venue's contract specification, funding history, margin rules, and settlement index—not a news summary alone:
41	```
42	get_financial_news(topic="BTC funding rate perpetual futures premium")
43	```
44
45	4. Stress all costs and failures: one-leg fill, reject, latency, transfer halt, depeg, funding flip, liquidation, borrow recall, and exchange default.
46
47	5. Report opportunity with: pair(s), spread %, estimated fees, net profit, and time sensitivity.
48
49	## Triangular Arbitrage Example
50
51	```
52	BTC/USDT: 50,000 \| ETH/USDT: 2,000 \| ETH/BTC: 0.041
53	Expected ETH/BTC: 2000/50000 = 0.040
54	Actual: 0.041 → ~2.5% spread (before fees)
55	Path: USDT → BTC → ETH → USDT
56	```
57
58	## Evidence and Validation
59
60	- Treat the setup as a testable hypothesis, not a prediction. Define thresholds, entry, invalidation, and exit before evaluating outcomes.
61	- Calibrate on the same instrument, venue, session, and timeframe. Use closed candles and a held-out or walk-forward sample; record every variant tried.
62	- Include spread, fees, slippage, borrow or funding, partial fills, and latency. Reject the setup when net expectancy is not positive or depends on one narrow parameter.
63	- Return observed inputs, missing data, cost assumptions, entry, invalidation, exit, and a valid, watch, or no-trade status.
64	- Research basis: [John, Li & Liu](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4816710) and [Vidal-Tomás](https://www.sciencedirect.com/science/article/pii/S154461232401393X) show that observed crypto spreads reflect liquidity, settlement, fees, and exchange-default risk—not risk-free profit.
65
66	## Key Rules
67
68	- NEVER report an opportunity without deducting all fees (trading, transfer, gas, slippage)
69	- NEVER assume execution is instant -- note that cross-exchange opportunities decay in seconds
70	- Pre-funded accounts on both sides are required for cross-exchange arb; transfers kill the edge
71	- Use a venue-, size-, and latency-specific minimum edge; no universal spread floor is valid
72	- Check order book depth: thin liquidity means slippage will eat the spread
73
74	## Related Skills
75
76	- funding-rate-trading -- funding rate arb (spot + perp) is a specific delta-neutral strategy
77	- altcoin-rotation -- cross-pair price analysis overlaps with rotation screening

People who liked this also viewed

6 picks

Type