The #1 mistake: it's not the coin, it's the size
Every crypto investor who "lost everything" made the same fundamental error: the position was too large relative to what they could actually afford to lose. The asset can be correct (it often is, in hindsight) and you can still be ruined if you can't stomach the volatility or if you owe obligations that the downswing wipes out.
So the first question is never "which coin?" It's: "how much of my net worth can I allocate to an asset that plausibly drops 80% and stays there for a year, without breaking me?"
Decide your size by asking: "If this entire allocation went to zero tomorrow, could I still comfortably meet every financial obligation for the next 12 months?" The answer is probably a smaller number than your gut wants. That gap between your gut and your actual capacity is where most blowups come from. Use the smaller number.
DCA: the unglamorous entry strategy that works
"DCA" = dollar cost averaging. Instead of making one big buy at some "perfect" price, you buy a fixed amount on a fixed schedule (weekly, monthly) regardless of price. Why this beats trying to time the bottom:
- You remove timing risk entirely. You can't catch the exact bottom — nobody can. DCA doesn't try to. It systematically acquires across prices, so your average cost is somewhere in the middle of the range, not at the top or bottom.
- You remove decision fatigue. No more checking the chart every day asking "is now the time?" The schedule makes the decision for you. This is the single biggest thing it does.
- You can actually stay in. A one-shot buyer at a local top, who then watches it drop 50%, is far more likely to panic-sell at the bottom. A DCA buyer has a lower average and a calmer relationship with the drawdown.
The one real cost of DCA: if the market just goes up in a straight line from day one, you underperform a lump-sum buy. That's the known tradeoff — and given that crypto reliably has deep drawdowns, most people will be better off DCA-ing than trying to one-shot the entry.
FUD and FOMO — the two enemies
These are not random. They're the two failure modes of a volatile, attention-driven market:
| FUD (Fear) | FOMO (Euphoria) | |
|---|---|---|
| When it hits | Drawdowns, red headlines, -30% day | Parabolic runs, green everywhere, "missing out" |
| What it makes you do | Buy at the bottom (sell low) | Buy at the top (buy high) |
| The real driver | Loss aversion + a position too large | Regret aversion + social proof |
| The defense | Pre-committed size + a written thesis you wrote in calm | A schedule (DCA) + a hard daily "no decision" window |
Write your thesis down in quiet, before you buy or hold: why this asset, what size, what would make you change your mind, and what your holding period is. When FUD or FOMO hits, don't argue with your emotions — open the document and re-read it. Your calm-self already made the decision. You're just remembering it.
Custody: where the coins actually live
Your security setup should match the amount: the more you hold, the less it should depend on any third party.
Covered by your process
- Small / active trading amount → CEX. Fine for the slice you're actively buying/selling. You accept that this is an IOU, not your coins.
- Moderate / long-term hold → self-custody software wallet. You control the keys. Back up the recovery phrase to at least two offline locations (metal backup for serious amounts is the standard advice because of fire/water/theft).
- Larger / multi-year → hardware wallet. A device that stores the keys offline. The most common "loss" in crypto isn't a hack of the network, it's a person losing their backup or getting socially engineered. A hardware wallet + a good backup dramatically reduces that.
Whatever the size, two rules never change: your recovery phrase is your account — never a screenshot, never a cloud note, and never shared. And test your recovery process once, in small, to make sure you can actually restore. A backup you've never verified is not a backup.
Which coins — a purpose-based filter
From Stage 1, the key distinction: each layer is a different risk. A simple filter before adding any "alt" to a portfolio:
- Is this the monetary bet? (Bitcoin) — store of value, macro-driven. Usually the core.
- Is this the platform bet? (Ethereum and L1s/L2s) — you're betting the ecosystem grows and the network succeeds. Higher risk, higher tech-dependence.
- Is this an app bet? (a specific DeFi protocol or token) — you're betting one product wins in a fast-changing category. This is speculative and the highest-risk layer. Size it accordingly.
"Not everything is created equal" isn't a slogan — it's a fact about the risk hierarchy. Treat the core (monetary) and the speculative (app tokens) as different asset classes with different sizing, not as a single "crypto" bucket.
The 90 / 10 barbell
The purpose-based filter above is a lens for deciding what each token is. A 90/10 split is a way of sizing them. It's the single most useful default to start from, and it comes straight out of the Everything Code: when one macro factor dominates, concentration is not reckless — diversification into a dozen low-conviction apps is what actually destroys returns.
SPECULATIVE 10% = apps, newer chains, higher-beta bets
| Sleeve | Size | What lives there | Why that size |
|---|---|---|---|
| Core | ~90% | Bitcoin (the monetary bet) + the major layer-1 networks (the platform bet) — e.g. Ethereum, Solana, SUI. | This is where the thesis lives. It should be large enough that a win actually matters, and stable enough that you don't have to watch it daily. |
| Speculative | ~10% | App tokens, newer chains, high-beta plays. The stuff that could be a 10x — or a 0. | Sized so a full loss is acceptable. This is the slot where curiosity and conviction are allowed to be loud, on purpose, at controlled risk. |
The 90/10 split is a starting point, not a law. The core should still fit the 100% loss test from earlier in this piece. And the 10% sleeve must be money whose total loss you can genuinely shrug at — the moment “speculative” starts to feel “necessary,” you have re-crossed the line. The split is there to make the speculative part a deliberate, budgeted choice instead of an emotional one.
Network effects: why the core is the core
The reason a 90/10 split is a defensible default — instead of spreading money across forty tokens — is not just “I like this one.” It is a structural fact about how assets with a network effect behave: they compound, they are hard to displace, and the winners accumulate a durable advantage even when a challenger ships a technically better product. This is the argument that makes “90% core” a rational bet instead of a superstitious one.
- The “winner takes most” shape. A network effect is a positive feedback loop. Every new user, developer or protocol on a network makes that network more valuable to the next user — so usage compounds rather than adding. In crypto, this is exactly why the top handful of networks by market cap have stayed there for years and why “the next coin to be 10x more useful” is not the same thing as “the next coin to be 10x the price.”
- Displacement is expensive. The practical corollary (and the one most useful to remember): you cannot out-feature a network with a real user base. You can match the features and win new users, but the incumbent collects the existing network — the developers, the integrations, the trust, the liquidity — as an ongoing head start. That is why the “better technology” argument is much weaker than it sounds for networks that already have scale.
- Reflexivity. There is a second loop, and this is the one that matters for a holder. Price movement draws in new users and developers; new users and developers drive real usage; real usage supports the price. This is the adoption curve, and it is why a network with a genuine base of users tends to recover faster from a drawdown than the same network without one. The reflexivity runs both ways — which is why the same dynamics that create parabolic runs create deep corrections.
- The value of the “build on here” layer. A network's total value is not just the native token. It is what builds on it: stablecoins, DeFi protocols, developer activity, settlement volume. When you look at a network's market cap against the total value that moves through it (stablecoins settled, DeFi TVL, transaction volume), the ratio tells you a more honest story than price alone. A network with a low market cap relative to its settlement volume is, all else being equal, undervalued against its peers.
It means the core is not “any coin I liked a few years ago.” It is the networks where the four signals above — scale, displacement cost, reflexive adoption, and build-on-value — are already measurable and pointing in one direction. When you hold a handful of networks on the Core side, you are not diversified out of habit — you are concentrated on the ones that have already won the network-effect argument. That is the whole reason the 90% is defensible at all, and the reason the 10% is where the real risk lives.
Reading on-chain data: the metrics that actually matter
On-chain data is the best kind of signal in crypto because it is independent of your opinion, mine, or anyone's marketing. But “on-chain data” is a misleading label for a field that is easy to misread. A few rules, in the order that matters most, from the way this kind of data is used in practice:
The four metrics worth checking, on a monthly basis at most, and in a fixed order so your read stays comparable over time:
| Metric | What it measures | Read it against |
|---|---|---|
| 1 · Total value settled | The real economic activity the network is doing — the dollars that moved, not the transactions that were counted. Hardest to fake because it costs real fees. | The same network's value settled last quarter, or the previous 90-day moving average. Upward trend with fee payment = genuine usage. |
| 2 · Fees paid (network-wide) | The total amount that users paid to use the network in a period. A direct measure of willingness-to-pay — people pay real money only for networks they actually trust enough to use. | Fee rate relative to the value settled. If fees are up but settlement is flat, the network is charging more without delivering more — a warning sign, not a bullish one. |
| 3 · Unique monthly active users | How many distinct people actually used the network this month. A more honest “is this real?” check than raw transaction counts, which can be inflated by bots and sybils. | Compare to the transaction count. If transactions are rising but active users are flat, that's concentration — the same users transacting more. That's not the same as growing adoption. |
| 4 · Stablecoin value on the network | How much stablecoin supply is settled or parked on the network directly. A direct proxy for real-world use — stablecoins exist to move value, so where they live tells you where real users are. | The same network's native-token market cap. Low market cap relative to stablecoin supply = undervalued against the activity it supports. High market cap relative to stablecoin supply = pricing speculation, not usage. |
Do not use these to time a trade. Use them to re-rank your core on a quarterly basis. “Is my current top pick still the strongest on the four metrics? If not, what is?” That question, asked calmly on a schedule, is what turns on-chain data from a dashboard into a discipline. When you combine it with the network-effect argument from the previous section, you have a real process for deciding which networks deserve the 90% — rather than assuming the answer on day one.
The full risk stack, in order
Next steps
This is educational process guidance, and not financial advice. Position sizing depends entirely on your personal financial situation, which no general article can know. Crypto can and does fall sharply; only invest what you can genuinely afford to lose. Sourced from Raoul Pal, Anthony Pompliano, Jordi Visser, Peter Diamandis & Bankless. The World is Spinning Faster Than Before is an educational project.