Smart money ยท All crypto, BTC & ETH, Major alts ยท Hours
Puffin
One Signal, All In
Senpi Signals already tells you where the proven cohort is positioned, where funding is dislocated and where open interest is surging.
- Smart money
- Aggressive risk
- Advanced
- Up to 10x
- From $50
- v1.2.2
- Proven traders
- Funding
- Open interest
How it works
Signals Hunter trades that feed across five names at a fifth of the wallet each. Puffin runs the identical engine and asks a harder question: out of everything scored in the last quarter hour, is there ONE name good enough to put the whole book behind? Most ticks the answer is no and it holds cash. That floor is set from measurement, not taste โ over 14 days of feed output, 82 of 116 scored observations clear Signals Hunter's 65 but only 39 clear 85, and since 21 of them sit at exactly 100 the score saturates, so 85 is close to the last level that still separates one signal from another. Know what the concentration costs before funding. It trades both directions, because the cohort is short about as often as it is long. When the cohort read fails it opens nothing and says so, rather than quietly scoring on price and funding alone and calling that a smart-money read.
Watches
Every liquid market in All crypto, BTC & ETH, Major alts
Every 15 minutes
Enters
Trades a published signal feed's own detector library, engine vendored verbatim.
Sizes
75% of its balance as margin
Up to 1 positions at once ยท up to 10x leverage
Exits
8-step trailing lock
None
The technical version
PUFFIN โ the same senpi-signals gather and detector library as SIGNALS-HUNTER, vendored byte-identically, but pointed at one position instead of five. A much higher score floor (85 vs 65), a single slot, most of the wallet behind it at 10x, and a 15-minute clock so the entry lands near the signal rather than up to an hour after it. Skips the tick outright when the proven-cohort lens is dark, because a score without it is OI/funding/price wearing the same name.
The idea
The detector library is the edge, so it is vendored rather than reimplemented: sweep.py, score.py and smartmoney.py are byte-identical to the skill's and a drift test enforces it, which means a detector fixed in the feed is fixed here. What this package changes is everything downstream of the score.
Read the full reasoning
One slot at 75% margin, 90% above a score of 93, at 10x โ up to 9x gross exposure on a single name. The scan clock drops from an hour to fifteen minutes, which buys entry timing rather than more trades: with one slot, turnover is set by the exit ladder and the cooldowns, not by how often the scanner looks. That faster clock forced a real fix rather than a config edit. score.py chooses both diff baselines by wall-clock age โ a roughly 60-minute fast arm and a roughly 12-hour trend arm that refuses below six โ so the inherited 24-deep ring, which spans a day at hourly ticks, spans only six hours at fifteen-minute ones and would have handed the trend arm a six-hour baseline while still printing it as twelve. The ring is kept stratified instead: dense inside two hours, then about one snapshot per two hours out to a day, which covers a longer span in fewer snapshots and gives both arms a properly aged partner. Exits are the DSL's trailing floor and nothing else. The ladder is Penguin's, which runs the same 10x, so each rung means the same price move in both: the first locks a floor above entry once the trade is up 0.6% in price, which cannot bank a winner but does stop one that went green from coming back through entry to the stop. The stop itself stays at 25% ROE, a 2.5% move in price, deliberately wider than Penguin's 1.5% โ a momentum jump is either right within the hour or wrong, while a rotation thesis is not falsified by a 1.5% move against it. One detector is still not traded: whale_open rests on a third-party startTime caught resetting on an unchanged position, which is a discountable sentence in a feed and the entire book here.
Where it can go wrong
- Proven traders can be wrong, early or hedged elsewhere. It sees their positions, not their reasons.
- At 10x, a 10% move against a position erases its margin.
- One position at a time, so each trade is a large share of the balance.
- No daily loss limit: it keeps trading through a bad day.
The rules
Version 1.2.2What it trades
- Markets
- Every liquid market in All crypto, BTC & ETH, Major alts
- Direction
- Long and short
- Positions at once
- Up to 1
- Typical hold
- Hours (intraday)
- Checks the market
- Every 15 minutes
Sizing
- Leverage
- Up to 10x
- Size per trade
- 75% of its balance as margin
- Minimum to start
- $50
Protection
- Profit lock
- 8 steps, from +6% on margin
- Daily loss limit
- None
- Drawdown halt
- Pauses after its balance falls 50% from its peak
- New trades a day
- No cap
- Losing streak
- After 3 losses in a row, waits 2 hours
8-step trailing lock. The first step keeps 30% of the peak once a trade is up 6% on margin (about 0.6% in price at 10x)
See the chart โThe profit lock
8 stepsAs a trade gains, the stop locks in a bigger share of its best level, and it only moves up. At +6% on margin it keeps 30% of the peak; by +100% it keeps 85%.
See all 8 steps
| When the trade is up on margin | Price move at 10x | The stop keeps |
|---|---|---|
| +6% | โ0.6% | 30% of the peak |
| +20% | โ2.0% | 40% of the peak |
| +25% | โ2.5% | 50% of the peak |
| +30% | โ3.0% | 60% of the peak |
| +40% | โ4.0% | 65% of the peak |
| +50% | โ5.0% | 70% of the peak |
| +70% | โ7.0% | 75% of the peak |
| +100% | โ10.0% | 85% of the peak |
Versions
2 versionsSame idea, tuned differently. Each is its own fork.
Make it yours
- Run it on gold and oil instead
- Cut the leverage to 5x
- Only take long trades
- Make the profit lock tighter
- Add a 10% daily loss limit
Your fork carries your name. Changing it never touches the original or anyone else's fork.
FAQs
What is the Puffin strategy?
Puffin is a smart money strategy for Hyperliquid. Senpi Signals already tells you where the proven cohort is positioned, where funding is dislocated and where open interest is surging.
What does Puffin trade?
Every liquid market in All crypto, BTC & ETH, Major alts. Long and short, up to 10x leverage, holding for hours.
How much do I need to start?
$50 is the minimum. You choose the amount when you fork it, and you can add more later.
Can I change how Puffin works?
Yes. A fork is your own copy: ask your agent to change the markets, leverage, sizing or exits and it rewrites the rules for you.
Why doesn't Senpi show Puffin's returns?
Every fork runs with its owner's money and settings, so no single return describes it. Backtests aren't shown either: they're fitted to the past. Start small and watch it trade.
Similar strategies
All smart money strategies
Take the read with you
The same read, refreshed on demand, in the Senpi Companion for Chrome, the app, and your own agent through Connectors.
Strategies are starting points, not advice. Trading perpetual futures with leverage can lose the money in a strategy.