CS2 Trade-Up Guide

CS2 Trade-Up Risk Management: Bankroll, Variance, and Stop Rules

Why can a positive-EV trade-up lose money?

Expected value averages every possible result by probability. Your actual contract produces one result. When a large share of EV comes from a rare high-priced output, many individual attempts can lose before that outcome appears—if it appears at all in your sample.

Risk management does not change the contract odds. It limits the damage when short-run results differ from the long-run average.

What numbers should you record before taking risk?

• Total basket cost.
• Net EV and expected ROI.
• Probability of finishing above break-even.
• Worst-case output proceeds and maximum dollar loss.
• Share of EV contributed by the best one or two outcomes.
• Output liquidity and likely time to sell.

If you have only ROI, you do not yet have a risk picture. Use our EV guide to calculate the full distribution.

How large should one contract be?

There is no universally safe percentage. Start from a hard dollar loss you can absorb without needing to sell other items or chase the result. Express the worst-case loss—not only input cost—as a percentage of the bankroll reserved for trade-ups.

Capital at risk = input cost − worst-case net proceeds
Risk fraction = capital at risk / trade-up bankroll

A $100 contract whose worst output nets $75 risks $25 under the modeled prices. But if that output is illiquid and might net only $55, the practical risk is $45. Use the conservative figure.

Why are several similar contracts correlated?

Contracts built from the same collections and sold into the same market share price risk. A Valve update, an input spike, or falling demand can hit all of them together. Running five recipes with different names does not create useful diversification if their valuable outcomes and liquidity depend on the same few skins.

How do you plan for a losing streak?

If a contract has a 40% profit chance, its loss chance is 60%. Assuming independent attempts and unchanged prices, the probability of five losses in a row is:

0.60^5 = 7.776%

That is uncommon, not impossible. The calculation also understates real-world risk when prices move between attempts. Choose size so a plausible streak does not force you to increase stakes, liquidate at bad prices, or abandon the process.

What stop rules prevent emotional decisions?

• Price stop: do not submit if refreshed net EV falls below your written threshold.
• Cost stop: abandon sourcing when the remaining basket cannot fit the maximum cost.
• Float stop: stop when available inputs cannot meet the maximum average normalized position.
• Loss stop: pause after a predetermined bankroll drawdown and audit assumptions.
• Liquidity stop: reject a contract when the modeled sale price lacks bids or comparable sales.

Define these before buying. A rule invented after a loss often becomes a justification for chasing it.

How should you track realized performance?

Keep a journal with the timestamp, selected inputs, exact floats, input cost, expected output distribution, modeled net EV, actual result, sale venue, fees, and realized proceeds. Separate model error from normal variance:

• Variance: the outcome was in the modeled pool but happened to be unfavorable.
• Model error: probability, wear, price, fee, or liquidity was wrong.
• Execution error: a wrong input, changed collection mix, or over-budget purchase altered the plan.

When should you not execute a trade-up?

Skip it when losing the at-risk amount would affect essential spending, when the output cannot be priced from credible market evidence, when one rare outcome supplies nearly all apparent EV, or when the basket requires rushing into unavailable floats. CS2 items are volatile digital goods, and Valve can change contract rules or market conditions.

Frequently Asked Questions