Why information traders need a bankroll framework
Information traders need a bankroll framework because being right is not enough. A good forecast can still lose money if sizing is reckless, liquidity is thin, or correlated positions cluster in one theme. The single biggest difference between profitable and unprofitable prediction market traders is not the quality of their forecasts — it is the discipline of their bankroll management.
This article walks through a complete bankroll framework: how to allocate capital into risk buckets, how to size individual positions, how to manage category-level exposure, and how to journal trades into a research process. It complements our risk management for traders playbook, which covers position-level discipline, and the forecast calibration playbook, which covers the probability-estimation layer underneath.
If you are new to the broader category, our introduction to prediction markets has the foundational mechanics.
Step 1 — Separate your bankroll into risk buckets
Different trades have different expected returns, variances, and time horizons. A bankroll that treats every trade as equivalent is a bankroll that gets pulled down by its riskiest positions and ignores its safest opportunities.
The standard three-bucket structure:
| Bucket | % of bankroll | Trade type | Expected hit rate |
|---|---|---|---|
| Core | 50–60% | High-conviction thesis trades with deep liquidity | 60–70% |
| Speculative | 20–30% | News-driven, faster-moving trades | 50–60% |
| Experimental | 10–15% | New categories, calibration learning, AI-assisted markets | 40–55% |
The core bucket is where your highest-conviction thesis trades live. These are positions where you have a clear edge, the market is liquid enough to support sizing, and the time horizon is long enough that intraday noise does not matter.
The speculative bucket is for news-driven and short-horizon trades. The hit rate is lower, but the trades are faster and the variance is bounded by tighter time stops. This is where most of the "interesting" trading happens, but it should never dominate your bankroll.
The experimental bucket is for learning. New categories, new market structures, new platforms, and AI-assisted markets all go here. The goal is not profit but calibration improvement — you are paying for data on whether your edge transfers to a new domain.
The cap on each bucket is non-negotiable. When you have a fantastic thesis but your core bucket is full, the right move is to wait — not to overload.
Step 2 — Translate edge into size with a tiered model
Pure Kelly is mathematically optimal but emotionally unsustainable. A tiered sizing model is easier to follow and produces similar long-run growth.
A practical model:
| Edge in basis points | Liquidity tier | Position size |
|---|---|---|
| 1–3 bps | Any | 0.25–0.5% bankroll |
| 3–7 bps | Tight spread + deep | 1–2% bankroll |
| 3–7 bps | Wide spread or thin | 0.5–1% bankroll |
| 7–15 bps | Deep | 2–4% bankroll |
| >15 bps | Any | Cap at 4–5% |
Notice the upper cap: even with a 20+ bps edge, never go beyond 5% on a single market. The reason is variance — a single market can dispute, the resolution can be delayed, the data source can change. No matter how confident you are, a 5% cap is the maximum reasonable single-market exposure for a retail bankroll.
The mathematical justification is the same as quarter Kelly. With a 5% per-market cap and a portfolio of 20 active positions, your bankroll experiences variance similar to quarter Kelly with the additional protection of explicit position bounds. The bounds matter because Kelly math assumes you know your edge — and you never do, exactly.
For the math behind these tiers, see the Kelly criterion section in our risk management playbook and the original Kelly 1956 paper.
Step 3 — Cap category and theme exposure
A 5% per-market cap is not enough. If you have four positions in a single category — say, four "interest rate cut" markets — you have a 20% bet on the same underlying thesis. One news event can wipe out the entire category at once.
The category cap structure:
| Category | Max % bankroll | Why |
|---|---|---|
| Macro | 25% | Highly correlated to Fed/policy news |
| Politics | 20% | Single events drive multiple markets |
| Crypto | 15% | High inherent volatility |
| Sports | 15% | Game-day correlation |
| Tech | 10% | News cycles compound quickly |
The category cap is enforced at the bankroll level, not the bucket level. If your core bucket is filling up on macro trades and your speculative bucket is also taking macro positions, the category cap means you cannot add another macro position until you trim.
Sophisticated traders also track theme caps within categories. "Risk-on" is a theme that spans macro and crypto. "Election volatility" is a theme that spans politics and crypto. Theme caps prevent you from accidentally double-counting your conviction across categories.
Step 4 — Build the trade journal that turns bets into research
A bankroll without a journal is a bankroll without learning. The journal is what turns isolated bets into a calibration process — the difference between a year of trading that taught you nothing and a year that materially improved your forecasting.
A minimum-viable trade journal:
| Field | Why it matters |
|---|---|
| Date | For time-series analysis |
| Market title | For grouping by category and theme |
| Category and theme | For exposure tracking |
| My probability | For calibration |
| Market price | For edge analysis |
| Edge in bps | For sizing audit |
| Position size | For risk tracking |
| Thesis | One sentence — what I believe and why |
| Exit plan | Price, time and event triggers |
| Resolution | Outcome and resolution date |
| P&L | Realized profit or loss |
| Lessons | What I learned, in plain language |
The lessons column is the most important. After 50 trades, group lessons by category. Patterns emerge: "I am overconfident in tech earnings calls", "I underweight the cost of wide spreads", "I sell winners too early in macro". Those patterns become the next quarter's calibration goals.
Public tools exist (Manifold's portfolio view, Polymarket's history API, Kalshi's reports), but a simple spreadsheet works fine. The discipline of writing matters more than the tool.
Step 5 — Review and rebalance quarterly
A bankroll framework that you set once and never review is a framework that decays. Markets change, your edge in different categories shifts, and the optimal bucket allocation evolves.
A quarterly review checklist:
- Calibration audit. Group last quarter's forecasts by probability bucket and check hit rates. Where are you overconfident? Underconfident?
- Category P&L. Which categories were profitable? Which lost? Is the edge real or did you get lucky?
- Exposure trends. Did you stay within category caps? Were there theme-level concentrations you missed?
- Bucket allocation review. Should you rebalance? Is your experimental bucket producing learning, or just losses?
- Process audit. Did you skip the journal on any trades? Did you trade without writing down edge?
The quarterly review is a meta-process. You are not optimizing individual trades; you are optimizing the framework that generates trades. This is where serious traders compound — not in the individual trades, but in the framework that makes the trades repeatable.
How a bankroll framework interacts with platform choice
Different prediction market platforms favor different bankroll structures.
Platforms with low fees and deep liquidity (e.g., the largest Kalshi and Polymarket markets) reward larger position sizing in the core bucket. Spreads are tight enough that even moderate edges survive execution.
Platforms with higher fees or thinner liquidity require smaller positions and a stricter edge threshold. The fee math we cover in market microstructure for forecasting is brutal at small sizes — a 2¢ edge on a 50 bps fee structure is barely profitable, while the same edge on a 200 bps structure is consistently losing.
Multi-platform traders should keep separate sub-bankrolls per platform. Withdrawal frictions, settlement times, and KYC limits all mean that capital on one platform is not freely interchangeable with capital on another. A 5% per-market cap is enforced per-platform, not across the entire combined bankroll.
For the math on platform choice, see comparing AMM liquidity models.
Common bankroll mistakes
Three patterns I see repeatedly.
Overweighting the speculative bucket. New traders find news-driven trades exciting and let the speculative bucket grow to 40–50% of bankroll. The hit rate is lower, the variance is higher, and the drawdowns are unsustainable. Keep speculative under 30%.
Skipping the journal. Every winning trader I know journals. Every losing trader I know skips it. The journal is the cheapest, easiest discipline to maintain, and it is the single biggest predictor of long-term profitability.
Treating bankroll as static. Your bankroll grows and shrinks. As it grows, your absolute position sizes grow with it — at quarter Kelly on a 10% edge, a $10k bankroll positions at $1k while a $50k bankroll positions at $5k. The percentage stays the same; the absolute dollar size changes. Conversely, in drawdown, you should be sizing in percentages off the smaller bankroll, not off your high-water mark.
Ignoring tax and withdrawal logistics. Realized P&L is taxable in most jurisdictions, and the tax timing affects how much of your bankroll is truly working. A 30% effective tax rate means that 70¢ of every $1 of realized profit is real bankroll; the other 30¢ is owed. Plan for this in your sizing, not at year-end when the bill arrives.
A worked bankroll example
A $20,000 bankroll structured for a serious retail trader:
- Core bucket: $12,000. Maximum 5 simultaneous positions, each capped at 5% of total bankroll ($1,000). Average position size $600.
- Speculative bucket: $5,000. Maximum 8 simultaneous positions, each capped at 3% of total bankroll ($600). Average position size $300.
- Experimental bucket: $3,000. Maximum 6 positions, each capped at 1% of total bankroll ($200). Average position size $150.
- Category caps: Macro $5,000, Politics $4,000, Crypto $3,000, Sports $3,000, Tech $2,000.
- Quarterly review: First Saturday of each quarter, 2-hour audit.
Total simultaneous positions: ~19. Total capital at risk at any time: ~$8,000. Reserve: $12,000 sitting idle, ready for high-conviction trades or drawdowns.
This structure produces a smooth equity curve over time and survives multiple bad weeks without psychological pressure. It is also boring — which is exactly the point.
Frequently Asked Questions
What is a bankroll framework and why is it different from a trading strategy?
A trading strategy describes how you identify and execute individual trades. A bankroll framework describes how you allocate capital across many trades, manage correlated exposures, and learn from outcomes. Strategy is about being right on one trade; bankroll framework is about being profitable across hundreds of trades.
How big should each risk bucket be?
The standard split is 50–60% core, 20–30% speculative, 10–15% experimental. The exact percentages depend on your risk tolerance and the quality of your edge in each category. Conservative traders push more into core; aggressive traders shift toward speculative. The experimental bucket should always be small enough that losses are tolerable as the cost of learning.
What is fractional Kelly and how is it related to bankroll frameworks?
Fractional Kelly is a position-sizing rule (typically 0.25× or 0.5× of full Kelly) that produces 50–75% of optimal growth with much less variance. A bankroll framework wraps fractional Kelly with bucket allocation, category caps, and journaling — it is the operational system around the sizing math. See our risk management playbook for the full Kelly derivation.
How often should I review my bankroll allocation?
Quarterly. Monthly is too frequent — you will react to noise. Annually is too slow — you will miss changes in market conditions or your own calibration. A quarterly 2-hour review is the right cadence for most retail traders.
Should I keep different bankrolls on different platforms?
Yes. Withdrawal frictions and KYC limits mean that capital is not freely interchangeable across platforms. Track each platform's bankroll separately and apply your sizing rules within each platform. The overall portfolio view is for category-level exposure tracking; the per-platform view is for sizing decisions.
How do I know if my framework is working?
Three indicators. First, your equity curve over rolling 6-month windows should be net positive. Second, your calibration in each probability bucket should be within ±5 percentage points of perfect. Third, you should be able to articulate, in writing, what you learned in the last quarter. If any of these fails, the framework needs adjustment.
What is the minimum bankroll to use this framework?
The framework scales down to about $2,000–$3,000 of bankroll. Below that, the per-market caps shrink to sizes where most platform fees consume the edge. A $500 bankroll with a 5% cap means $25 positions, which on most platforms is below the practical fee threshold. Build the bankroll to $2k+ before applying the framework rigorously.
Can I use this framework for cross-platform arbitrage?
Yes, with modifications. Arbitrage trades have very different risk profiles — they are usually lower-edge, higher-volume, and faster. Most arbitrage traders use a separate sub-bankroll within the experimental bucket, with custom sizing rules that prioritize execution speed over edge size. The journaling discipline still applies.
Where to go next
You now have a complete bankroll framework. The natural follow-ups:
- For the position-level discipline that lives inside the framework, see risk management for prediction market traders.
- For the calibration loop that makes your edge estimates honest, see the forecast calibration playbook.
- For the microstructure that determines whether your edge survives execution, see market microstructure for forecasting.
- For platform choice considerations, see comparing AMM liquidity models.
A bankroll framework turns trading from a series of isolated bets into a research process. The traders who internalize that distinction are the ones still in the market five years from now.
