Dieser Workshop ist auch auf Deutsch verfügbar.
Eight modules, each with a lab. At the top of every lab you will find missions; they tick themselves off once you have solved them. You need no prior knowledge beyond percentages.
Module 1 of 8
Futures
A derivative is a contract whose value depends on another price. The simplest one is the future: fix the price today, buy or sell later.
- Long and short The buyer is long, the seller is short. Payment happens at expiry.
- Margin Instead of the full price you post collateral, often 5 to 20%. That is where leverage comes from.
- Mark-to-market Gains and losses are settled against your account as they happen.
- Basis The gap between future and spot. Annualised, it is your return in a cash and carry.
Try this
- Slide the future to its fair value.
- Make the future cheap enough that the reverse trade pays.
- Play two price paths and compare the total at expiry.
Cash and carry: buy spot, sell the future. You earn 0.71 per 100 more than the capital costs you.
Dry run: random price, fixed result
Day 90 · price 134.49
Example values, not market data. Fair value S · e^(rT); simple annualised basis. The dry run uses 60% volatility; the future converges to spot by expiry.
Why F = S · e^(rT)?
You need bitcoin in a year. Route one: buy it today for 100 and hold it; that ties up capital that would otherwise earn 4% interest. Route two: buy a future. Both routes have to cost the same, or there is an arbitrage. So the future sits at about 104.
- : spot price
- : interest rate p.a. (financing cost)
- : time to expiry in years
This is not a model of the future but an arbitrage argument.
Contango, backwardation and settlement
A future above spot is in contango; below spot, in backwardation. In crypto, contango is the norm: many traders want leveraged long exposure and pay a premium for it.
At expiry the underlying is either delivered (physical) or only the difference is paid (cash settlement). Crypto futures on CME and Deribit settle in cash.
Module 2 of 8
Options
An option is a right, not an obligation: you only buy or sell if it pays. That right costs a premium.
- Call The right to buy. Pays when the price ends above the strike.
- Put The right to sell. Pays when the price ends below the strike.
- Strike The price at which you may buy or sell.
- Premium What the right costs. The buyer can lose at most this.
Try this
- Drag the strike K right on the chart.
- Find the strategy that wins in a crash and in a rally.
- Let time run down to expiry.
- +1 Call K 105
Pay a premium, bet on rising prices. You cannot lose more than the premium.
Premiums from Black-Scholes with a zero rate. One contract refers to one unit of the underlying here.
European, American and moneyness
European options can only be exercised at expiry, American ones at any time. Most crypto options, for example on Deribit, are European.
A call with strike 90 at a price of 100 is in the money, with strike 110 out of the money, with strike 100 at the money.
Module 3 of 8
Pricing
In 1973 Fischer Black and Myron Scholes showed how to rebuild an option from the underlying and cash. What the rebuild costs is the fair price.
Try this
- Get the call's delta to 0.50.
- Show gamma and shorten expiry to one week.
- Drag S or K on the chart.
| Metric | Call | Put |
|---|---|---|
| 7.01 | 6.68 | |
| 0.542 | −0.458 | |
| 0.0231 | 0.0231 | |
| 0.114 | 0.114 | |
| −0.119 | −0.108 |
Put-call parity: both sides always match.
The formula
- : cumulative distribution function of the standard normal
- : volatility p.a.
That is why option traders often talk about vol rather than prices: "the 30-day call trades at 55 vol."
The Greeks at a glance
| Greek | Measures | Rule of thumb |
|---|---|---|
| Delta | Price change per 1 in the underlying | ATM call ≈ 0.5 |
| Gamma | Change in delta per 1 in the underlying | largest ATM and close to expiry |
| Vega | Price change per vol point | longer expiries react more |
| Theta | Price change per day | buyers lose time value every day |
Not every option has a formula. Then you simulate: roll many price paths, average the payoffs, discount.
Try this
- Run a simulation.
- Simulate again: how much does the estimate wobble?
- Set the strike to 130 or higher and simulate.
Terminal prices from geometric Brownian motion, rate 4%. The estimate is the mean of the discounted payoffs; the funnel shows the error shrinking with 1/√N.
Module 4 of 8
Perpetuals
A perp is a future that never expires, and today the most traded product in crypto. With no expiry to pull it to the spot price, another mechanism takes over: funding.
- Funding Longs and shorts pay each other a small amount every few hours. The direction pulls the perp towards spot.
- Premium How far the perp trades above or below spot.
- Leverage At leverage L you post only 1/L of the notional.
- Liquidation If the price moves roughly 1/L against you, the exchange closes your position.
Try this
- Find the zone where funding is exactly 0.01%.
- Make the shorts pay the longs.
- Push funding above 100% p.a.
The perp trades above spot. Longs pay shorts; being long gets more expensive and the price is pushed back.
Rule as on Binance and BitMEX: funding = P + clamp(0.01% − P, ±0.05%). With hourly payments each hour pays one eighth of the 8-hour rate.
High leverage is normal on perps, 50× and more. The stress test shows what that means.
Try this
- Get liquidated once.
- Find a leverage with under 10% risk over 30 days.
- Run five stress tests and compare with the model.
Start a run: one random price path over 30 days.
Rule of thumb for an isolated position without fees. The probability assumes normally distributed returns; real crypto returns have fatter tails, so the number is more likely too low.
The funding formula
- : premium index, the averaged gap between perp and spot
- : interest component, usually 0.01% per 8 hours
- : limited to the range −a to +a
Binance and BitMEX pay BTC funding every 8 hours, Hyperliquid every hour (one eighth of the 8-hour rate each time). According to the BitMEX derivatives report for Q3 2025 BTC funding was positive more than 92% of the time.
Liquidation price and auto-deleveraging
- : leverage
- : maintenance margin, e.g. 0.5%
If the margin does not cover a liquidation, the exchange's insurance fund steps in. If that is empty, profitable positions on the other side are cut automatically (auto-deleveraging). It hits exactly the traders who are winning.
Where perps come from
The idea goes back to a 1993 paper by Robert Shiller. It took off in 2016 when BitMEX launched the first bitcoin perp. Today perps trade more volume than the spot market.
Module 5 of 8
Crypto markets
The maths is the same as in traditional markets. What differs is the market structure, and that is where the opportunities come from.
Try this
- Open all six quirks.
Always open
No closing bell, no holidays. Liquidity is thinner at night and at weekends; moves are often larger then.
Many decentralised exchanges have no order book but a pool with one rule: the product of the reserves stays constant, . The price follows from their ratio.
Try this
- Sell enough that the price impact goes above 5%.
- Let the arbitrageur push the pool back.
- Set the pool fee to 0.
The pool now quotes below the exchange price of $3,500. An arbitrageur buys 19.97 ETH back from the pool, sells it on the exchange and keeps $1,368.51.
Constant-product pool as in Uniswap v2. The arbitrage profit ignores gas and the fee on the way back.
More on AMMs
As the Uniswap v2 whitepaper describes, the gap after each trade is by design: traders' fees pay liquidity providers for the risk that arbitrageurs profit at their expense. The gap is often closed within the same block.
Module 6 of 8
Arbitrage
In an arbitrage the profit is almost fixed at entry. It is rarely free: you pay with capital, speed or a risk that does not appear in the model.
The classic in crypto is the funding carry: long spot, short perp, same notional. Price moves cancel out and the funding is what remains.
Try this
- Find a holding period where fees eat all the funding.
- Push the return above 15% p.a.
- Switch to simulated funding and roll a new path.
The short leg gets liquidated after a rise of roughly 32.8%. The spot side gains just as much, but it sits in a different account; without a top-up the hedge is gone.
Simplified model: funding on notional, no price moves, liquidation at 0.5% maintenance margin. The simulated path is made up and not a forecast.
Faster and smaller are price gaps between exchanges, or between three pairs on one exchange.
Try this
- Find the spread at which trading across two exchanges pays.
- Find a profitable triangle route that starts with BTC.
- And one that starts with ETH.
No trade: the spread does not cover the costs.
Example values: BTC $100,000, ETH $3,500. In practice gaps like these close within milliseconds, and slippage eats part of them.
How cross-exchange arbitrage works in practice
Nobody moves coins back and forth for it; that takes too long. You hold inventory on both exchanges and rebalance now and then. Those who collect these gaps reliably have fast connections, low fee tiers and code that decides within milliseconds.
Module 7 of 8
Strategies
Nine strategies on one map: on the right the profit is fixed at entry, on the left there is only a statistical edge. At the top speed matters, at the bottom patience and capital.
Try this
- Compare three strategies on the map.
- Look at all nine.
Funding carry
Long spot, short perp, same notional. Price moves cancel out and the funding is what remains. The income varies; you can exit at any time.
- You earn
- Funding on the short side
- Main risk
- Negative funding, liquidation, ADL
Module 8 of 8
Check
One question per module. The explanation comes right after your answer.
Try this
- Answer all eight questions.
- Get at least six right.
Question 1 of 8
Spot is at 100, the rate is 4% p.a. Roughly where should a one-year future trade?
0 of 0 missions across the workshop
Where to go next
The classic on derivatives is "Options, Futures, and Other Derivatives" by John Hull. For perps, the exchanges' documentation is worth reading, for example Hyperliquid's explanation of funding.
You learn most by rebuilding a lab yourself, as a notebook with real data. That is what our meetings are for: open to every subject and experience level.