← Back to dashboard

Learning to read TIDE

The short version — everything you need to get started, nothing you don't.

What is this about?

Imagine checking whether tomatoes are cheap or expensive at the market today. That's what TIDE does — for four cryptocurrencies (BTC, ETH, SOL, HYPE) — by boiling lots of small clues into one easy number.

TIDE never tells you to buy or sell. It just shows you what the numbers say right now. The decision is always yours.

Why TIDE works the way it does

A pre-registered study of this system has been running since 1 September. The hypotheses, the methods and the rules for judging them were written down before the first number existed — with a timestamp, so they cannot quietly be adjusted afterwards. The results get published either way, including the uncomfortable ones.

The data for it builds up around the clock. Every 15 minutes a background system records 20 individual factors for each coin and each time frame — roughly 16,000 snapshots in a fortnight. Whether or not anyone has this page open.

The rules stay unchanged while that happens. Adjusting them after a good run or a bad one is the fastest way to mistake luck for skill.

And what TIDE can't do is written down too — further down this page, not in the small print.

Tide and Pulse — the two big ideas

Tide

Like low and high tide at the beach. Low tide, the beach is open — easy pickings. High tide, everything's underwater — harder. Tide shows if a price looks cheap or expensive right now.

Pulse

Like at the doctor's — a racing pulse means stress, a calm one means ease. Pulse shows whether big traders using borrowed money are feeling bold or cautious.

The big circle — the only number that really matters

Each coin gets a colored circle with a number, 0 to 100. Read it like a traffic light:

ColorNumberMeaning
● Green70–100Looks cheap right now
● Cyan40–69No clear direction
● Red0–39Looks expensive right now

That's the whole idea. Everything else on the page just explains why the number looks the way it does.

The wavy line under the circle

Every minute or two, TIDE writes down the current score — 45, 47, 52, 58, 61... Instead of showing you that list of numbers, it just draws them as a little line: low number, low point on the line; high number, high point.

So this line isn't the price again — you already have that in the chart next to it. It's showing how TIDE's own opinion has been moving over the last several minutes. Line going up means TIDE has been getting more convinced. Going down means the opposite.

One thing that trips people up: the line always stretches to use its full height, even for a tiny wobble. If the score only moved between 61 and 64, the line still swings from top to bottom — because within that short window, 61 was the lowest and 64 was the highest it got. Read the shape (rising, falling, flat) rather than how tall the wiggle looks.

Right when you open the page or switch time frame, you'll briefly see a plain, evenly-repeating wave instead — just a placeholder until at least two real readings exist, so it never pretends to show data before there's any.

Three buttons: which time frame?

Think binoculars with different zoom levels — Day Trading zooms in close (the last few hours), Swing is medium (days to weeks), Hodl zooms all the way out (months to years, the big picture).

The same coin can be green under Day Trading and red under Hodl at once. Not a bug — just two different distances on the same thing.

One honest thing worth knowing about Hodl specifically: it deliberately thinks like a value investor, not a trend-follower. "Cheap" mostly means far from the all-time high, in a fearful market, after a long cooldown — the exact opposite of "strongly rising, greedy, and hitting new highs," even when that rise is genuinely healthy. That means Hodl can read "expensive, be cautious" during a strong, real uptrend that a pure trend-following chart would call bullish. Both views are legitimate and both are used by real professionals — they're just answering different questions ("is this cheap relative to its own history" vs. "is the trend intact right now"). Hodl is built to ask the first one, on purpose.

The chart and the one sentence below it

Next to the circle is a small candle chart — mostly green means the price has been climbing, mostly red means it's been falling. You don't need to read it in detail.

Right under it is one sentence labeled "Auto insight" — a plain-language reason for why the circle looks the way it does. Short on time? That sentence alone is usually enough.

Want to know more? Tap the ☰ menu

Setup levels — turning the circle into concrete numbers

When the circle is clearly green or clearly red, a box called "Setup levels" appears underneath it. It has a few parts — here's each one in plain terms.

Entry band

Not one exact price, but a small range, like "$81.40 – $83.10". Think of it like a parking zone rather than one exact parking spot: a few different signals happen to point at roughly the same area, so instead of pretending to know the one perfect number, TIDE shows the whole zone where they agree.

Next to it, a small tag reads Tight, Medium, or Wide — how narrow that zone actually is compared to the coin's normal daily wobble. Tight means the signals landed close together, worth more trust. Wide means they only roughly agree — treat it as a looser guess. Occasionally you'll also see a note like "includes a real order book wall ($2.1M)" — that means a genuinely large stack of real buy or sell orders happens to sit inside this same zone, not just technical math pointing there.

Confluence

A percentage, like "78%", showing how many of the 20 clues actually agree with this direction. Higher means more of them point the same way — like counting how many friends agree with a plan before you commit to it. Below 50%, the number itself dims slightly and gets a small "thin" tag right next to it — TIDE flagging its own weak spot rather than letting a shaky number look as confident as a strong one.

You'll sometimes see this as "11/20" and other times as "9/11" — that's not a mistake. It only counts clues that actually have a real say in the score right now. More on exactly why, just below in The sliders.

Stop-loss ladder

Not one stop line, but three — like three fences of increasing size:

  • Soft — the closest one, easily crossed by normal daily wobble.
  • Hard — a bit further out, a more meaningful break.
  • Structural — the outermost one. If price gets here, the whole picture this circle was built on has changed.

Tap the little arrow to see the exact price of each fence. A note like "2/3 intact" tells you how many haven't been crossed yet. Sometimes a small "crowded positioning" warning shows up next to the outer fence — it means a lot of borrowed-money traders are already betting the same way (see Pulse above), and crowded bets can unwind faster than usual.

Reward-to-risk range

The same "is the potential reward worth the risk" math, shown for all three time frames (Day, Swing, Hodl) side by side — so you can see how it shifts depending on how far you zoom out.

The track-record line

Occasionally a small line appears saying something like "67% reached Target 1". That's TIDE keeping an honest scoreboard of its own past setups, right in your own browser — a real count, never a guess. Most of the time this line won't show up at all, simply because not enough setups have actually finished playing out yet on your device. That's on purpose: TIDE would rather show nothing than make up a number.

When the circle sits in the cyan middle zone, there's no directional box — instead you'll see two "trigger" prices, one for a possible long and one for a possible short. These just mark the nearest real technical levels, not a prediction of which way price will go — more like "here's what would need to happen for either direction to become clearer."

How a professional trader would read each piece

Same numbers, but here's the quick, no-nonsense way experienced traders tend to size them up:

The one habit that ties all of this together: never judge a setup from a single number alone. A professional looks at the whole set together — band, confluence, ladder, and reward-to-risk — before deciding anything means something.

The banner: "Clearest signal right now"

Sometimes a small banner appears above the four coins, naming one of them by name. It's answering one specific question: out of all four coins, and all three time frames, which single setup currently looks the most convincing — long or short, doesn't matter?

"Convincing" here means distance from the middle of the scale, not a high score. A very low score (strong short case) counts exactly as much as a very high one (strong long case) — 50 is the only value that means nothing either way.

What a setup has to clear

But a big score swing alone isn't enough to earn the banner. It only appears once the exact same checks a professional would apply are ALL satisfied at once: confluence of at least 50%, a stop-loss ladder that's still mostly intact, and a reward-to-risk of at least roughly 1.5:1, measured to Target 2 — the level a win is counted at — except on Day, where the bar sits a bit lower (around 1.2:1), since a day trade's naturally tighter price range makes the same 1.5:1 harder to reach even for a genuinely good setup. Most of the time this banner shows nothing at all — that's not a bug, it means nothing right now clears all three bars simultaneously.

One more check runs quietly behind these: a "cheap, buy it" call gets blocked while that same time frame's own trend is clearly still falling, and an "expensive, short it" call gets blocked while it's clearly still climbing — the classic way a contrarian read gets dangerous is firing straight into a trend that hasn't actually turned yet. The one exception is a genuine extreme: history keeps showing that real capitulation and euphoria moments are exactly when a contrarian call is most often right, even while the trend itself still looks intact in the "wrong" direction — so a strong enough reading is allowed through regardless. A second, more specific way through this same exception exists too: even a moderate reading gets through if the actual signature of a real washout is present — price stretched a couple of ATRs from its own average, the crowded side of the market visibly paying up to stay in (via funding), and no fresh positions still piling in on that side.

The search runs continuously across Day, Swing, and Hodl together, even the two you aren't currently looking at — it happens quietly in the background, so a strong Swing setup can still show up while you're looking at Day, without you ever needing to click over and check yourself.

The quality tag

Once a setup clears those three bars, it also earns a small quality tag next to the coin's name:

  • Good — cleared the minimum bar. Tradeable, nothing more claimed.
  • Very good — reward-to-risk, confluence, and the stop-loss ladder are all meaningfully strong at once, and the score itself isn't sitting right at the edge of neutral either — shown with a soft amber glow around the banner.
  • Excellent — the same three, but at their strongest, plus a score far from neutral, plus genuine agreement spread across different kinds of evidence rather than several closely-related clues all making the same basic point — shown with a brighter, gently pulsing gold glow.

One great number by itself can never lift a setup into a higher tier — reward-to-risk, confluence, and the ladder all have to be strong together, so a single standout figure can't hide two mediocre ones.

When several qualify at once

If more than one setup clears the bar at once, the banner doesn't pick just one and ignore the rest — it quietly rotates through every qualifying one, every few seconds, with small dots showing how many there are and which one you're looking at right now. Only the very first slide is genuinely "the clearest" — each one after it is labeled honestly by rank ("Signal 2 of 3") instead of every slide claiming the same top spot.

Since each slide carries its own rating, the glow around the banner changes with it too — don't be surprised to see it shift from a soft amber to a brighter, pulsing gold as it rotates from a "Very good" setup to an "Excellent" one, or lose the glow entirely on a plain "Good" one. And whichever slide happens to be showing is also the one tapping the banner jumps you to — if it's currently showing a Swing setup, tapping takes you to Swing, even if the very first slide was a Day one.

Small arrows beside the dots let you step through by hand. Tapping one pauses the rotation for a few seconds afterward, so your own tap doesn't get immediately undone by the next automatic switch — handy on a phone, where there's no mouse to simply hover and pause. On a computer, hovering over the banner pauses it for as long as your pointer stays there.

Track record — did those signals actually work out?

Every time a signal clears every quality gate, TIDE writes it down — which coin, which time frame, the full entry zone, both targets, and all three stop levels. A background system then checks the real price every 15 minutes and fills in what actually happened, independently of whether this page is even open.

You'll find this near the bottom of the page, under "Track record". It starts closed to keep the page tidy — tap it to open, and you'll see every signal ever saved, whether it hit a target, got stopped out, or is still waiting.

One honest thing worth knowing before you trust the numbers: a handful of resolved signals doesn't prove much yet. This becomes meaningful once dozens have played out across genuinely different market conditions, not after the first two or three.

Why you might not see many signals yet

TIDE is currently in a deliberate freeze phase — the scoring rules and thresholds stay exactly as they are until real signals have played out across a few genuinely different market situations, not just more of the same one. Changing the rules too early, based on too little evidence, is exactly how a tool ends up chasing noise instead of real patterns.

That doesn't mean nothing is happening behind the scenes. A handful of extra, purely observational measurements run quietly on every check in the meantime. They never affect a score, a gate, or a live decision — they simply build up a real body of evidence, so that any future change is based on what actually happened, not a guess made on day one.

What these numbers don't tell you

Every tool has limits. Most don't say what theirs are. Here are TIDE's, as of 18 September 2026 — written down so you don't have to find them out the hard way.

Resolved on 18 September 2026: the reference R:R and the track record used to describe two different trades. Until then, the reference R:R assumed a limit entry inside the zone while the track record measured an entry at the market price at the moment of saving — the reference figure could be several times the measured one. Since the rule change, both describe the same trade: a limit entry at the edge of the zone, measured to Target 2. Signals from before that date stay in the record as they were measured, marked by their rule version. The setup still shows how far the current price sits from the entry.

A timeout can be a signal that was doing fine. When a signal is resolved, Target 2, the stop the gate was measured against, the time limit — and, since 18 September 2026, Target 1 — are checked. Since that date, touching Target 1 moves the stop to the entry: a signal that reached Target 1 and then came back ends as "breakeven" (0 R), not as a loss, and one that reached Target 1 and then ran out of time is marked "timed out (touched T1)". Neither counts as a win — only Target 2 does. The track record also shows how often Target 1 was touched at all.

A hit rate means nothing without its chance level. For a market with no predictability at all, the probability of reaching a target before a stop is simply stop distance ÷ (target distance + stop distance). With a stop 4% away and a target 2% away, a coin flip hits 67% of the time. TIDE's first 78 signals touched Target 1 in 74% of cases — and the chance level for their geometry was 66%. That is not skill; that is geometry. Since 19 September 2026 the track record shows the chance level next to every rate, and TIDE only claims skill where the rate sits clearly above it.

The track record runs on 17 of 20 factors. The background system that fills in results cannot reach Binance positioning, CVD, or order-book imbalance. Those three contribute nothing to the recorded score. The live dashboard in your browser uses all twenty — so the number you see live and the number in the track record can differ by a few points.

The entry band can look tighter than it is. A band is called "tight" when the price references that form it sit close together. If only a single reference is found, the band has no width at all — and comes out as "tight" through the absence of other references rather than through their agreement.

Day and Swing read the same candles. Both styles are built from four-hour candles from the same data source; they differ in how far back they look and in their indicator settings, not in the raw material. So when Swing appears to "confirm" Day, that is a second reading of the same data, not an independent second opinion.

All of these are recorded in the project's decision log, together with what will be done about them and when. None of them will be quietly fixed and forgotten.

A few more things you might spot

The sliders — and only the ones turned up actually get a vote

Under "Weighting" you can re-mix those 20 clues yourself — don't care about market mood? Slide it to 0. Made a mess? There's a reset button right next to the heading.

Why the count changes

Here's the part that explains that "11/20" vs. "9/11" thing from Confluence above: think back to the friends-agreeing-on-a-plan idea. If three of your friends couldn't make it to the meeting, would you still count their opinion in the final tally? Of course not — you'd only count the ones actually in the room. TIDE works exactly the same way. A clue sitting at 0% weight — whether because you slid it there, because this time frame naturally ignores it (Hodl doesn't care about a fast intraday tool like MACD), or because its data just isn't available right now — isn't "in the room" for this decision. It never gets counted in either half of the confluence number.

Hodl is a good real example. Nine of the twenty clues sit at 0% there by design (MACD, Bollinger, OBV, VWAP, ATR, CVD, Order Book Imbalance, Liquidation Magnet and Volatility Percentile — all fast, short-horizon tools that have nothing useful to say about a multi-month view). So on Hodl, only 11 clues are ever "in the room":

agrees disagrees 0% weight — not counted

9 agree out of 11 that actually count → 82%, not 9/20.

The total you see next to the percentage always reflects exactly this — however many clues genuinely have a say for this coin, this time frame, and your own slider settings, right now.

Starting out: just leave them as they are. They're already set sensibly.

See the exact weight every clue gets, side by side across all three styles

All 20 clues at a glance

One line each — the full explanation for any of these is one tap away in the ☰ menu. Trend-following clues (marked *) read genuine outperformance as bullish, the opposite convention from the contrarian ones around them — both are legitimate, just different questions.

ClueDaySwingHodl
RSI — overbought or oversold right now?895
Trend (EMA) — above or below its own average?8109
MACD — is the trend speeding up or slowing down?580
Bollinger %B — near its statistical ceiling or floor?760
OBV — does volume confirm the price move?660
VWAP — cheap or expensive vs. today's average?900
ATR — is short-term volatility expanding or calming?540
Fibonacci — where in its recent range is price sitting?3106
ATH distance — how far from the all-time high?0424
Fear & Greed — how anxious or greedy is the whole market?0520
Momentum — has it run far over the longer term?0015
Binance Top Trader — are big leveraged traders long or short?585
CVD — are active traders buying or selling more aggressively?830
Futures Basis — trading at a notable premium or discount?453
Order Book Imbalance — more real buy or sell orders nearby?820
OI/Volume — are positions being held or churned quickly?656
Funding Extremity — unusually high or low funding for this coin?233
Liquidation Magnet — does a nearby cluster pull price up or down?640
Relative Strength* — beating or lagging the other three coins?654
Volatility Percentile — historically calm or extreme for this coin?430

Every column adds up to 100. A 0 doesn't mean "broken" — it means that clue's own time horizon simply doesn't apply here (VWAP resets daily, so a year-long Hodl position gets nothing useful from it; Momentum's multi-month lens has nothing to say about the next 30 minutes).

The most important thing

TIDE helps you understand numbers. It is not an advisor telling you what to do. Nobody can predict the future. The decision is always yours — ideally after doing your own research too.

Small dictionary

TermPlain explanation
PriceWhat one coin costs right now
ScoreThe 0–100 number in the circle
ZoneWhether the score reads cheap, neutral, or expensive
Trading styleYour chosen time frame: Day Trading, Swing, or Hodl
Order bookThe list of everyone currently waiting to buy or sell
This guide belongs to the TIDE dashboard at kryptodashboard.org. For anything not covered here, the ☰ menu inside the dashboard explains every term with its own picture, right next to the live numbers.