8.8 AI for Crypto Traders

AI won't hand you a money printer — anyone selling "AI signals" with a rented Lambo in the thumbnail is selling the thumbnail. What AI actually is: the hardest-working unpaid analyst in history — research, summaries, code, and devil's advocacy on tap. Used right, it compresses hours of homework into minutes. Used wrong, it confidently agrees you should ape into a coin named after a frog.

Using AI to summarise news

Feed it articles, threads, or a week of headlines and demand structure:

Summarise these articles for a crypto swing trader. Format:
1) Facts only (no opinions) 2) What's actually new vs. recycled
3) Bullish/bearish/neutral for BTC and ETH specifically, with one-line reasoning
4) What the articles are NOT saying that I should check myself
Keep it under 200 words. If sources contradict each other, say so.

Point 4 is the pro move — AI is excellent at spotting the dog that didn't bark.

Building research prompts

The quality gap between "is X coin good?" and a real research prompt is the whole game:

Act as a sceptical crypto analyst reviewing [COIN]. Cover:
1) What does it actually DO, in one sentence a bogan would accept?
2) Tokenomics: supply schedule, unlock cliffs coming, who holds the top 10 wallets?
3) Team: doxxed? Track record? Anything they'd rather I didn't find?
4) The three strongest reasons this goes to zero (steelman the bear case)
5) What would need to be TRUE for this to 10x — and how likely is each condition?
Flag anything you're uncertain about instead of guessing.

Portfolio analysis

Paste your holdings (coins and percentages — no account numbers, no keys, obviously) and ask for a stress test: "What's my real diversification if these correlate 0.9 in a crash? What's my drawdown if BTC drops 40%? Which single position is quietly carrying my whole risk budget?" AI is superb at the "mate, your 'diversified portfolio' is eleven flavours of the same bet" conversation your friends are too polite to have.

Strategy testing & coding assistance

This is the biggest unlock — module 8.7 with the difficulty dial turned down:

• "Write a Pine Script v6 strategy: long when 20 EMA crosses above 50 EMA and price is above the 200 EMA; 2× ATR stop; exit on cross-down; include commission of 0.1%." — working backtest in a minute.
• Paste your bot's error message and the offending function — AI debugging is genuinely faster than forums ever were.
• "Review this bot code for risk holes: what happens if the API times out mid-order? If the same signal fires twice? If the exchange returns a partial fill?" — AI as code reviewer catches the 3am scenarios beginners never imagine.

Market sentiment analysis

Feed it raw crowd noise — a hundred headlines, a subreddit thread, Fear & Greed readings — and ask for a temperature check with a contrarian lens: "Is this euphoria, despair, or boredom? What would a contrarian note?" Remember the 8.3 lesson: peak retail excitement is distribution weather. When AI says the crowd is unanimously bullish, that's not confirmation — that's a weather warning.

Creating custom indicators

Describe any indicator idea in plain English — "an indicator that paints the background green only when ADX > 25 AND price is above VWAP AND the 20 EMA slope is positive" — and get working Pine Script. Iterate conversationally ("make the ADX threshold an input", "add an alert condition"). Your 8.5 three-gauge dashboard can become ONE custom indicator with one signal light. That's not cheating; that's tooling.

Risk analysis — AI as the designated driver

The highest-value habit in this module. Before any significant trade, run the devil's advocate prompt:

I'm about to [LONG/SHORT] [COIN] at [price], stop at [X], target at [Y],
risking [Z]% of my account. My reasoning: [paste your actual reasoning].
Attack this trade. What am I not seeing? What's the strongest case I'm wrong?
Check my R:R maths. Is my stop at an obvious liquidity level (would I be
easy to hunt)? Rate the trade plan 1-10 against textbook risk management
and tell me what a disciplined trader would change.

You're not asking permission — you're renting a second opinion with no ego and no FOMO. If your trade can't survive an AI poking it for thirty seconds, it definitely can't survive the market.

8.9 Risk Management — The Module That Keeps You Out of the Trolley

If this book were on fire and you could save one module, save this one. Every strategy in 8.4 loses sometimes; every indicator in 8.5 lies sometimes. Risk management is the only thing standing between "losing trade" and "losing trader". It is the seatbelt, the rollcage, AND the airbag.

Position sizing — the only formula that matters

You never decide "how much to buy". You decide how much you're willing to LOSE, and the maths tells you the size:

Notice what this kills: "I'll just put five grand on it" (size chosen by vibe), and "the stop's too far so I'll skip the stop" (the last words of a thousand accounts).

Maximum drawdown — the arithmetic of holes

Losses and gains aren't symmetrical. The hole is always easier to dig than to climb out of:

This is why pros obsess over drawdown, not returns. The 1–2% risk rule exists because of this table: at 1% risk, even ten straight losses — which WILL happen to every trader eventually — is a 10% hole, not a funeral.

Risk/reward ratios — revision with teeth

From 8.1: minimum 1:2 on every trade, so a 40% win rate still pays. Here's the compounding kicker — combine the sizing formula with 1:2 minimum and a 45% win rate, and you have a complete profitable system with NO prediction skill at all. Everything else in this book — the candles, the SMC, the indicators — exists to nudge that 45% toward 55%. The risk maths is the engine; analysis is just the paint.

Diversification & portfolio allocation

• Real diversification is rare in crypto. When BTC sneezes, 95% of altcoins catch pneumonia — in a crash, correlations go to 1 and your "ten different coins" become one big red candle wearing ten hats.
• The Boganster allocation for traders: the majority of crypto wealth in majors, held in the hardware wallet from Chapter 1 (untouched, untraded, unleveraged) — and a trading float of maybe 10–20% of the crypto stack in the arena. The float is the only money the strategies of 8.4 ever touch. Blow up the float and it's a bruise; the stack is fireproof because it's not in the building.
• Max 3 concurrent positions while learning, never all in correlated alts, and position #2 in the same narrative (two AI coins, two dog coins) counts as adding to position #1 — because in a dump, it is.

Daily loss limits — the circuit breaker

Pre-commit, in writing, in the plan: lose 3% of the float in a day (or three trades, whichever first) and you are DONE for the day. Not "one more to win it back" — that's the tilt talking, and tilt has the worst win rate of any strategy ever devised. The best traders' superpower isn't picking winners; it's being incapable of having a catastrophic day. Your bot gets the same rule hard-coded (8.7); you get it written on a sticky note on the monitor. The market opens again tomorrow. It always does.

The weekly review process

1. Read every journal entry from the week. Ten minutes. No skipping the embarrassing ones — especially not the embarrassing ones. That's where the tuition is.
2. Sort trades into four buckets: good process/won, good process/lost (fine!), bad process/lost (expected), and the sneaky killer — bad process/WON. That last bucket is poison in a winner's jersey: it teaches you that breaking rules pays. Flag every one.
3. Find the week's repeated leak. There's almost always one — early exits, oversized "conviction" trades, trading the chop. Name it out loud like it's at a meeting: "I'm Dave, and I move my stops."
4. Set ONE fix for next week. One. Measured. ("Zero stop adjustments this week" — countable, binary, done or not.)
5. Check the account maths: current drawdown, average R:R, win rate, risk-per-trade drift. Five numbers, five minutes, no lying.

8.10 Professional Resources — The Full Toolshed

Every tool a professional setup needs, with an honest one-liner on what each is actually FOR. You don't need all of these — you need one from most categories, chosen once, learned well. (Some course links are affiliate/referral links, as ever: cost you nothing, keep the School's lights on.)

Exchanges

Hardware wallets (Chapter 1 rules apply: official store only, never second-hand, ever)

Software wallets (spending money only — the stack lives on hardware)

Portfolio tracking

On-chain analysis (reading the blockchain's actual books)

Market data

Token security (run EVERY unfamiliar token through one of these BEFORE buying)

Whale tracking

Economic calendar (strategy 11's bible — check it every Monday, no exceptions)

AI tools (module 8.8's crew)

Crypto news

Podcasts (turn the commute into tuition)

YouTube channels

Bonus Downloads — The Glovebox

Every tool referenced in this chapter, ready to grab from the members' area. Built once, properly, so you don't have to.

• Trading journal (Excel/Google Sheets) — the 8.1 journal pre-built: entry/exit/R:R columns, emotion score, auto-calculated stats, and the four-bucket review from 8.9.
• Position size calculator — the 8.9 formula as a spreadsheet: account, risk %, entry, stop in; size out. No mental arithmetic at 2am.
• Risk calculator — drawdown scenarios, win-rate/R:R breakeven table, "how many losses until I'm at X%" modelling.
• Crypto tax spreadsheet — trade-by-trade log formatted to hand straight to your accountant (who would like a word about last year, by the way).
• Portfolio tracker — stack vs. float split, allocation percentages, correlation warnings.
• AI prompt library — every 8.8 prompt plus twenty more, copy-paste ready.
• Trading checklist — the four-point pre-entry gate from 8.1, printable, laminate-able, stick-it-to-the-monitor-able.
• Daily routine checklist — the pre-flight and post-session routines as tick-boxes.
• Weekly review template — the five-step 8.9 review as a fill-in form.
• Monthly performance review — the bigger-picture audit: equity curve, rule adherence, one-metric-to-fix.
• Bot starter code — the 8.7 RSI skeleton plus templates for all six projects, testnet-configured, keys-in-env, risk limits included.
• Pine Script examples — the 8.6 scripts plus every 8.4 mechanical strategy as a ready-to-backtest strategy() file.
• Strategy cheat sheets — each 8.4 playbook on one printable page: rules, conditions, mistakes.

Practical Assignments — where it becomes real

Reading this chapter loaded the ute. The assignments drive it. In order, no skipping — each one builds on the last:

1. Build your TradingView workspace. Layouts (Bias + Execution), your three-gauge indicator template, alerts at this week's key BTC/ETH levels. Done when: you can open one saved layout and know the market state in 60 seconds.
2. Build a 20-coin watchlist — sorted into Majors/Radar/Zoo, with one written sentence per coin on why it's there. "Shazza mentioned it" is not a sentence that survives this exercise. Done when: every coin has a reason you'd defend out loud.
3. Write your one-page trading plan — style, setup checklist, risk numbers, session times, review schedule (the 8.1 template). Done when: a stranger could trade your plan without asking you a single question.
4. Paper trade ONE strategy for 20 trades — same strategy, every signal, full journal, no real money. Done when: 20 trades are journalled and you know your win rate, average R:R, and your one repeated mistake — because you WILL find one, and finding it free of charge is the whole point.
5. Build the RSI bot on an exchange testnet — the 8.7 skeleton, running on fake money, logging every decision for two weeks. Done when: it has survived a fortnight unattended and the log explains every action it took.
6. Keep the journal for 30 consecutive days — every trade, every emotion score, plus the weekly reviews. Consecutive means consecutive; day 22 with no entry restarts the clock, same as the gym. Done when: the habit no longer requires willpower.
7. Write your performance review — three biggest strengths, three areas to improve, with journal evidence for all six. Then pick ONE improvement for next month. Done when: you've read it out loud to someone — the dog counts, but the mirror is braver.