Copy Size Conviction Trader
This is a template. The default signal uses portfolio concentration as a proxy for whale conviction. When a top trader commits a large fraction of their capital to a single market, that is a stronger signal than raw position size or trade count. The skill handles all plumbing (leaderboard scraping, activity reconstruction, concentration analysis, signal boosting, safeguards). Your agent provides the alpha by tuning concentration thresholds.
Strategy Overview
Not all whale positions are equal. A trader who puts $100K into a market might be spreading $5M across 50 markets (low conviction) or concentrating $300K total into 3 markets (high conviction). This skill distinguishes between the two.
- Discover top wallets from the predicting.top leaderboard
- Reconstruct each wallet's portfolio from on-chain trade activity
- Measure portfolio concentration: position_cost / total_portfolio_cost per market
- Filter positions where concentration exceeds MIN_CONCENTRATION threshold
- Boost trade sizing proportional to whale conviction level
- Validate every trade against conviction-based YES/NO threshold bands
The Edge: Concentration as Conviction
Most copytrading tools treat all whale positions equally. A $50K position from a $5M whale (1% concentration) gets the same weight as a $50K position from a $150K whale (33% concentration). The second whale is clearly more convicted.
Signal Logic
Leaderboard (top N wallets)
|
v
For each wallet:
Fetch 200 most recent trades
Reconstruct open positions (buys - sells per market)
Compute concentration = position_cost / total_portfolio_cost
|
v
Across all wallets:
Find markets where any whale has concentration >= MIN_CONCENTRATION
Aggregate: market -> [(wallet, side, concentration, cost), ...]
Sort by max concentration descending
|
v
For each high-conviction market:
Match to Simmer market catalog
Run compute_signal() for base conviction (YES/NO threshold bands)
Verify whale side aligns with signal direction
Apply concentration multiplier:
conc_mult = 1 + min(CONCENTRATION_MULT - 1, concentration / 0.5)
whale_boost = min(0.3, (num_whales - 1) * 0.1) [if multiple whales agree]
boosted_conviction = min(1.0, base_conviction * conc_mult + whale_boost)
size = max(MIN_TRADE, round(boosted_conviction * MAX_POSITION, 2))
|
v
Execute or paper-trade
Why This Works
- Concentration is observable: Unlike sentiment or conviction surveys, portfolio concentration is computed directly from on-chain data. It cannot be faked without moving real capital.
- Asymmetric information: Whales who concentrate are betting they know something. A 30% portfolio allocation implies the whale expects significant edge in that specific market.
- Multi-whale convergence: When multiple independent whales concentrate in the same market, the signal compounds. This is harder to game than a single whale's activity.
- Conviction alignment: The skill only trades when whale concentration aligns with the base YES/NO threshold signal, preventing blind copying into neutral or overpriced markets.
Example Dashboard
| Market | Whale | Conc | Side | Size |
|---|---|---|---|---|
| Will X happen by June 2026? | 0xabc12... | 32% | yes | $4,200 |
| Federal rate cut in Q3? | 0xdef34... | 28% | no | $8,100 |
| Federal rate cut in Q3? | 0x789ab... | 15% | no | $2,900 |
| Bitcoin above $150K by July? | 0xcde56... | 22% | yes | $6,500 |
| Major tech IPO in 2026? | 0xfab78... | 11% | yes | $1,800 |
In this example, "Federal rate cut in Q3?" has two whales with high concentration on the same side -- the multi-whale boost kicks in for stronger conviction sizing.
Safety
| Guard | Default | What it does |
|---|---|---|
| Paper mode | ON | No --live flag = simulated trades, zero risk |
| MAX_POSITION | $50 | Cap per-trade size |
| MIN_TRADE | $5 | Floor prevents trivially small orders |
| MAX_POSITIONS | 8 | Portfolio-level position limit |
| MAX_SPREAD | 10% | Skip illiquid markets |
| MIN_DAYS | 5 | Skip markets resolving too soon |
| MIN_VOLUME | $3,000 | Skip thin markets |
| YES/NO thresholds | 0.38/0.62 | Only trade within conviction bands |
| MIN_CONCENTRATION | 10% | Ignore positions below this portfolio share |
| Side alignment | ON | Whale side must match signal direction |
| context_ok() | ON | Prevents exceeding max open positions |
Tunables
All parameters are configurable via SIMMER_* environment variables and the Simmer UI:
| Env Var | Default | Description |
|---|---|---|
SIMMER_MAX_POSITION |
50 | Max position size in USD |
SIMMER_MIN_TRADE |
5 | Min trade size in USD |
SIMMER_MIN_VOLUME |
3000 | Min market volume in USD |
SIMMER_MAX_SPREAD |
0.10 | Max bid-ask spread |
SIMMER_MIN_DAYS |
5 | Min days until resolution |
SIMMER_MAX_POSITIONS |
8 | Max open positions |
SIMMER_YES_THRESHOLD |
0.38 | YES threshold for conviction bands |
SIMMER_NO_THRESHOLD |
0.62 | NO threshold for conviction bands |
SIMMER_MIN_CONCENTRATION |
0.10 | Min portfolio concentration to qualify |
SIMMER_CONCENTRATION_MULT |
2.0 | Max multiplier from concentration |
SIMMER_LEADERBOARD_LIMIT |
15 | Top N wallets to analyze |
Remix Ideas
- Sector concentration: Only count concentration within a category (crypto, politics, sports) to find domain-specialist whales
- Concentration velocity: Track how concentration changes over time -- a whale increasing from 10% to 30% is actively building conviction
- Inverse concentration: Fade whales who are decreasing concentration (selling down), as this signals loss of conviction
- Concentration clustering: Find markets where 3+ whales all have >15% concentration -- crowd conviction premium
- Time-weighted concentration: Weight recent positions more heavily than older ones when computing portfolio share
- Whale-relative sizing: Instead of a fixed multiplier, size proportional to the whale's actual dollar concentration
Dependency
Requires simmer-sdk (pip install simmer-sdk) and a valid SIMMER_API_KEY.