q&a¶
┌─┐ questions worth answering once. │?│ answers worth re-reading. └─┘
a curated FAQ across every section of the guide. each answer is short, opinionated, and links you back to the page where the topic is treated in depth.
01 / getting started¶
I'm completely new — where do I begin?
Read 01-foundations/ end-to-end first, in the order the section index lists. Don't jump ahead. Most people who blow up early skipped the foundations because they thought they were "obvious." They're not — they're the substrate everything else rests on. Then 02-mathematics/statistics-basics. Then pick one strategy in 05-quantitative-strategies/ and study it for a month before opening a brokerage account.
How long until I should be trading real money?
Months, not weeks. Paper-trade a defined system for at least 4–8 weeks with the exact rules written down. If your written rules survive contact with reality, deploy with the smallest position size your broker allows. Scale up only after you have meaningful sample data on your real-money execution. See position sizing for the math.
What's the difference between trading and investing?
Trading is short-horizon and bet-driven — you take a position because you have a view that resolves over seconds to months. Investing is long-horizon and ownership-driven — you hold an asset because you believe its fundamental value will compound over years. Same instruments, different intent. Most beginners are temperamentally suited to investing and force themselves into trading because trading looks faster. It isn't, on a risk-adjusted basis.
Do I need a degree in finance or math?
No. You need fluency in basic statistics (mean, variance, hypothesis testing), comfort with Python (or R, or Julia), and willingness to read primary sources. A CS or physics degree is usually more useful than an MBA. Many of the best traders are self-taught.
Should I learn TA, FA, or quant first?
Quant — the meta-skill of testing a hypothesis on data — generalizes to all three. Once you can backtest, you can evaluate any technical or fundamental claim. Without that, you're forced to take what others tell you on faith.
Is this guide enough, or do I need books too?
This guide is a structured map. Books fill in depth. Start with: Trading and Exchanges (Harris) for microstructure, Active Portfolio Management (Grinold & Kahn) for quant, Reminiscences of a Stock Operator (Lefèvre) for psychology. After 18 months, Advances in Financial Machine Learning (López de Prado).
What's a realistic timeline to consistent profitability?
For most people who get there at all: 2–5 years of dedicated study, paper trading, and small live trading. Most people quit in year one because results don't appear. The ones who succeed are the ones who treat the first 2 years as a tuition payment, not a salary expectation.
Can I trade part-time alongside a day job?
Yes — swing or position trading is well-suited to people with day jobs. Day trading and scalping are not, full stop. The market doesn't accommodate your schedule, and divided attention is fatal at short timeframes.
How do I know if trading is even right for me?
Honest answers to: (1) Can you take a loss without it affecting your mood for the rest of the day? (2) Can you follow a rule you wrote yesterday even when you "feel" differently today? (3) Are you willing to be wrong publicly and update? If any answer is no, build the skill before you scale up. See 11-psychology.
02 / capital & costs¶
How much capital do I need to start?
The math says ≥ $25,000 to make the per-trade economics work in US equities (the PDT rule also kicks in below this for day trading on margin). With less, you can still trade — but commissions, spread, and slippage become a larger fraction of each trade's edge. For crypto and futures, much less can work for small experimental sizing. Frame the question as: can my edge per trade exceed total round-trip cost?
What costs am I actually paying per trade?
More than the headline commission. Real cost ≈ commission + bid-ask spread + slippage + market impact (large orders) + financing (overnight margin) + opportunity cost (capital tied up). On liquid US equities, total round-trip cost can be 5–15 bps. On illiquid micro-caps or crypto alts, 50–500 bps. Always model costs explicitly in your backtesting engine.
Is leverage worth using as a beginner?
No. Leverage amplifies both edge and mistakes — and beginners' mistakes are larger than their edge. Trade unleveraged until you have ≥ 6 months of profitable live results, then introduce small leverage cautiously, sized via Kelly or half-Kelly.
How much should I risk per trade?
1–2% of account equity per trade as a starting heuristic. The exact number depends on win rate and reward:risk — see position sizing for the formal derivation. Risking more than 2% per trade means a 10-trade losing streak (a routine event) can wipe out 20%+ of your account.
What's the cheapest way to test a strategy?
Paper trading on a broker that simulates real fills (slippage, partial fills, rejections). Most brokers offer this free. Avoid backtest-only validation — backtests systematically overstate live performance by 30–50% in our experience and across most published studies.
Do I need a Bloomberg terminal?
No, unless you're working at a buy-side firm. For retail / sub-$10M AUM use, public data providers (Yahoo Finance, Alpha Vantage, Polygon, CoinGecko, Federal Reserve FRED) plus a few inexpensive paid sources cover 95%+ of strategies. Bloomberg starts paying for itself at institutional scale — large fixed-income, options, and credit desks specifically.
03 / strategy selection¶
Should I pick momentum, mean-reversion, or carry?
All three work in different regimes. Most retail traders pick one based on personality fit. Momentum suits patient trend-followers who can sit through chop. Mean-reversion suits contrarians comfortable with frequent small losses. Carry suits people who can stomach long quiet periods punctuated by sharp drawdowns. A portfolio across all three is more robust than any single one — see factor investing.
How do I know if a strategy still works?
Track three things in real time: (1) rolling 3-month Sharpe ratio vs backtest expectation, (2) win rate, (3) realized vs expected drawdown. If any drifts meaningfully outside the backtest's confidence interval for 3+ months, the strategy may be in a different regime. See regime detection.
Is technical analysis 'real' or just superstition?
Most TA is genuine pattern-recognition with weak statistical edge after costs. Some elements (volume confirmation, multi-timeframe alignment, support/resistance) have measurable backtest evidence. Other elements (Elliott waves, candlestick magic) don't survive serious testing. Treat TA as one input among many, not as a self-contained system. See 03-technical-analysis.
Are options strategies easier than equity trading?
No — options add 4 new risk dimensions (Greeks) on top of directional risk. They give you more tools but more ways to be wrong. Start with covered calls and cash-secured puts; only progress to multi-leg structures once you can compute and rebalance Greeks on a daily basis.
Is HFT or algo trading realistic for an individual?
HFT requires co-location, custom hardware, and 7-figure infrastructure budgets — not realistic for retail. Algorithmic trading at slower horizons (intraday to daily) is very realistic and is what most retail quants actually do. See 10-trading-systems for the architecture.
Should I copy successful traders?
No, unless you can replicate their exact rules including risk management. "Copy trading" platforms expose you to the trader's whole portfolio including bets you don't see being placed. Survivor bias is severe — for every visible "successful" trader there are 50 invisible ones who blew up.
What's the best timeframe for a beginner?
End-of-day or swing (1-5 day holds). Slow enough to let you think between decisions, fast enough to give meaningful sample size in a year. Day trading and scalping require execution skills and emotional control that take years to build.
04 / risk management¶
What's the single most important risk metric?
Maximum drawdown — both historical and expected. Sharpe ratio is what you market; max drawdown is what you live through. A strategy with Sharpe 2.0 and 40% drawdown is a strategy you won't actually stick to during the drawdown. See drawdown management.
Should I use stop losses?
For directional discretionary trades — yes, always, defined before entry. For mean-reverting strategies — stops can actually hurt you (they exit at the worst price). For options spreads with defined max loss — you don't need stops on the underlying, the spread structure is your stop. Match the stop methodology to the strategy. See stop-loss strategies.
How do I size positions correctly?
Use the Kelly criterion or a fraction of it (half-Kelly is the practical default). Don't use "intuition" — the math is well-defined and small sizing errors compound brutally over time.
What about position sizing for correlated trades?
Treat correlated positions as one position. Five long tech stocks at 1% each is closer to 5% concentrated tech risk than 5% diversified equity risk. Either correlation-adjust the sizing, or constrain total exposure to a correlated bucket. See correlation analysis.
How do I think about tail risk?
Assume the empirical worst-case in your data is not actually the worst case. Markets routinely produce events worse than anything in a 20-year sample. Use stress tests with hypothetical scenarios, hold some always-on tail protection (small OTM puts, long-vol), and never allow a single tail event to exceed your maximum acceptable drawdown. See tail risk protection.
Should I diversify across asset classes?
Yes, but only if you actually understand the asset classes you're adding. Adding crypto or commodities to "diversify" without understanding their dynamics is just adding new ways to lose money. Real diversification = low-correlation return streams + skill in each. See 09-asset-classes.
What's a 'kill switch' and do I need one?
A pre-defined daily / weekly / monthly loss limit that, when hit, forces you to stop trading until you review what happened. Yes, you need one — written in advance, ideally enforced by your software, not by your willpower. The most expensive trades are the ones placed in the hour after a big loss.
05 / tools & platforms¶
What programming language should I use?
Python for nearly everything — data analysis, backtesting, ML, API integration. It has the largest ecosystem (pandas, numpy, scipy, statsmodels, scikit-learn, PyTorch). R for academic statistics. Julia for performance-critical numerical work. C++ only if you're doing latency-sensitive HFT.
Best free tools to start with?
Python + pandas + Jupyter for analysis. TradingView for charting. Yahoo Finance or Alpha Vantage for free data. Interactive Brokers or Alpaca for low-cost execution (API-friendly). MkDocs + git for documenting your own strategies. Total cost: $0.
Open source backtesting frameworks?
backtesting.py (simple, vectorized) for prototyping. vectorbt for large-scale parameter sweeps. zipline-reloaded or nautilus_trader for event-driven simulation closer to live behavior. Build your own if your strategy doesn't fit any of these. See backtesting frameworks.
Data sources — paid vs free?
Free is fine for backtesting and education on liquid US/EU equities, futures, and crypto. Paid (Polygon, Tiingo, IEX Cloud, Refinitiv) becomes worth it for: tick-level data, point-in-time fundamentals (no survivor bias), options chains, less-liquid assets. Start free, upgrade only when a specific strategy needs it.
Cloud vs local?
Local is fine for backtesting and most live trading at retail scale. Cloud becomes useful when: (1) you need 24/7 uptime (crypto, FX), (2) latency to exchanges matters more than your home internet provides, (3) you're collaborating. AWS Lightsail / DigitalOcean droplets are usually enough — you don't need k8s.
How do I store and version my strategies?
Git. Each strategy is a repo or directory. Track config, code, backtest results, live performance logs. Code review your own changes before deploying to live — even if you're a solo trader, the discipline matters.
06 / backtesting & validation¶
Why does my backtest look so much better than live?
Almost certainly some combination of: look-ahead bias, survivorship bias, optimistic execution assumptions, no slippage, no commissions, overfitting via parameter search, or selection bias from running many backtests and keeping the best. The honest live Sharpe is usually 30–60% of backtest Sharpe. See common pitfalls in backtesting engine.
How many trades do I need to validate a strategy?
Rule of thumb: 100+ trades for a weak conclusion, 300+ for moderate confidence, 1000+ for high confidence. Below 30 trades, you can't statistically distinguish skill from luck regardless of how good the headline metrics look.
What's a 'walk-forward' backtest and why does it matter?
Re-fitting the strategy on a rolling window of historical data and testing on the next out-of-sample window. Repeats across time so you simulate "what would this strategy have looked like if I had been re-fitting it monthly the whole way?" Catches strategies that only work with a single hand-picked parameter set.
How do I avoid overfitting?
(1) Hold out 20–30% of data untouched until your strategy is finalized. (2) Cap the number of parameters at the minimum needed. (3) Test robustness — does the result hold across slightly different parameter values? (4) Apply walk-forward validation. (5) If you tested 50 ideas to find one that works, apply Bonferroni correction before believing the result.
Should I trust a Sharpe of 3+?
Be very skeptical. Real, live, after-cost daily Sharpes above 2 are rare even at top hedge funds over long periods. A backtest showing 3+ usually has data leakage, look-ahead bias, or insufficient costs. Investigate before deploying.
What about transaction costs?
Always model them, even crudely. A common approximation: commission + half-spread + a slippage term scaling with volume × notional. If your strategy only works at zero costs, it doesn't work.
07 / psychology & discipline¶
Why do I keep deviating from my own rules?
Because following them when they hurt feels worse than breaking them in the moment. The fix isn't more willpower — it's making deviation harder (kill switches, broker-side limits, position-size caps coded in software). See emotional discipline.
How do I handle a big loss psychologically?
Stop trading for the rest of the day, minimum. Document what happened in your journal — facts, not emotions. Sleep on it. The single most expensive class of trades is the "revenge trade" placed in the hour after a loss. The market will be there tomorrow.
Is journaling worth the time?
Yes, and most people quit too early. The point is not to review every single trade but to spot patterns across hundreds. After 6 months of consistent journaling, your own pattern of mistakes becomes the most actionable input for improvement. See journaling.
How do I know if a losing streak is variance or a broken strategy?
Compare to backtest expected drawdown at the same trade count. A 10-trade losing streak when your backtest expected one every 200 trades is variance. A 10-trade losing streak with win rate dropping 15+ percentage points below baseline is more concerning. If drift persists 3+ months → reduce size, re-validate the strategy on recent data.
How do I avoid revenge trading?
Pre-commitment: a hard daily loss limit enforced by either software or a person (spouse, accountability partner). Once hit, no more trades that day. Period. Most major wipeouts trace to one bad afternoon where someone "tried to make it back."
Do I need to be unemotional?
No. You need to be able to follow your rules despite emotion. Emotion isn't the enemy; acting on emotion mid-trade is. The same emotion (fear of loss) that hurts you mid-position can help you with sizing pre-position.
08 / tax & regulatory¶
Are short-term and long-term trades taxed differently?
In most jurisdictions, yes. In the US, positions held < 1 year are taxed as short-term (ordinary income rates, can be 37%+). Positions held > 1 year qualify for long-term capital gains (0–20%). This dramatically affects after-tax returns. A strategy with high turnover needs much higher pre-tax Sharpe to match a low-turnover strategy.
Do I need to register as a trader?
In the US, "Trader in Securities" status is a tax election with specific requirements (substantial activity, regularity, intent to profit short-term). It allows mark-to-market accounting and ordinary loss treatment but requires diligent record-keeping. Talk to a tax professional before electing. Other jurisdictions have similar concepts (UK CGT vs trading, Australian PSI, etc.).
Crypto tax — what's the simple version?
Every crypto-to-crypto trade is a taxable event in most jurisdictions (not just crypto-to-fiat). Track every transaction: date, asset in, asset out, cost basis, fair market value in fiat at the time. Use a crypto tax tool (Koinly, CoinTracker) — manual tracking is hopeless past ~50 trades.
Do I need to report futures and options separately?
In the US, Section 1256 contracts (most futures, some options) get blended 60/40 long-term/short-term tax treatment regardless of holding period — usually favorable. Equity options are taxed by holding period like stocks. See tax implications.
Is algorithmic trading regulated?
For retail, mostly not — you're using your broker's API which the broker has registered. For institutional algo trading (e.g., MiFID II algorithmic trading definition in the EU, SEC Rule 15c3-5 in the US for sponsored access), there are notification, testing, and kill-switch requirements. See compliance.
09 / crypto specifics¶
Is crypto a viable trading market or just gambling?
Both. Spot BTC/ETH and major-cap alts are tradeable with the same disciplines as any other asset class. Long-tail tokens (low market cap, low liquidity) are closer to gambling — adverse selection is severe. Stick to top-50-by-market-cap and treat the rest as venture-style speculation, not trading. See crypto.
What's different about trading crypto vs equities?
24/7 markets, no circuit breakers, much higher volatility (3–5×), thinner order books in alts, exchange counterparty risk, custody risk, regulatory uncertainty, and an inverted leverage effect — positive returns predict higher future vol, opposite of equities. Position size accordingly.
Should I self-custody?
Yes for any meaningful holdings you're not actively trading. Exchange failures (Mt. Gox, FTX, Celsius) have happened repeatedly. "Not your keys, not your coins" is a hard-learned cliché. Hardware wallets (Ledger, Trezor) for the bulk; small float on exchange for active trading.
Are stablecoins safe?
Define "safe." USDC and USDT have proven liquid through multiple stress events but both have had brief de-pegs. Algorithmic stablecoins (e.g. UST) have collapsed. Treat stablecoins as bank deposits at the issuer, not as money. Diversify across stablecoins for material balances. See stablecoin flow signal.
Funding rate arbitrage — is it really free money?
No. The funding rate carries the basis risk between perp and spot, which can move violently in stress. It also carries exchange counterparty risk, and during liquidation cascades the arb breaks down precisely when you need it to work. Real returns net of these risks are positive but much smaller than headline funding yields. See funding rate arbitrage.
DeFi vs CeFi for execution?
DeFi (Uniswap, dYdX, GMX, etc.) for self-custody and on-chain trades. CeFi (Coinbase, Binance, Kraken) for tighter spreads and faster execution. Most active traders use both — CeFi for hot float, DeFi for self-custodied size and access to specific markets (perp DEXes, on-chain perps). See market structure.
10 / career & learning path¶
Can I make a living trading my own account?
Possible but rare. A realistic threshold: ~\(500K–\)1M in capital, 15%+ net annual returns sustained over 3+ years. Below that, the math (living expenses ÷ realistic returns) doesn't work. Most people who try this route either fail, or supplement with consulting / teaching / writing income.
Should I work at a hedge fund or trade my own money?
Different trade-offs. A hedge fund gives you scale, infrastructure, data, mentorship, and risk capacity you can't replicate solo — at the cost of equity in your own work. Trading solo gives you full ownership and flexibility — at the cost of slower learning, no peer pressure, and full personal downside. For most people, 3–5 years at a fund first, then optionally going solo, is the right sequence.
What skills do prop firms / hedge funds hire for?
Quant research roles: probability/statistics, ML, programming (Python + C++ for some), domain knowledge in your asset class, ability to write clearly. Discretionary trader roles: track record, articulating the process behind decisions, ability to size and hold under stress. Engineering roles: low-latency systems, distributed systems, market-data pipelines, exchange protocols.
How do I build a track record people will take seriously?
Trade live (not paper) with at least $10K for at least 18 months, with audited statements. Write up your process, the strategies, the wins and the losses honestly. A 2-year audited record at modest size beats a 10-year backtest at any size.
Best path: CFA, MFE, PhD, or self-study?
Depends on the role. CFA — useful for buy-side asset management, less so for trading. MFE (Master of Financial Engineering) — strong for quant roles at banks and mid-tier funds. PhD (physics, math, CS, stats) — opens doors at top quant funds (Renaissance, Two Sigma, D.E. Shaw). Self-study — works if you have the discipline and can demonstrate output (open-source projects, papers, audited live track record).
Didn't find your question?¶
The full guide is structured to answer most questions in depth. Try the search bar (top right) or browse by section from the home page.
If you think a question belongs here but isn't — open an issue or PR. The bar: questions whose answers are useful to multiple people and stable over time (not "what does the market do tomorrow").