NydarTrading

Posted on Feb 24 • Originally published at nydar.co.uk

We Just Built an Institutional-Grade Order Flow Suite From Scratch

#webdev #trading #fintech #beginners

The Problem With Retail Trading Tools

Most retail platforms give you a price chart, a basic order book, and maybe a volume bar at the bottom. That is like trying to navigate a city with a compass and no map.

Professional traders at prop desks and hedge funds see something completely different. They see the mechanics underneath price — where large orders are hiding, which levels are being defended, where leveraged positions will be forced to liquidate. They see the tape, the footprint, the delta, the full depth across multiple exchanges simultaneously.

That toolset has historically cost $2,000-5,000 per month. Dedicated order flow platforms charge serious money because the data and the rendering are genuinely hard engineering problems.

We just built all of it. From scratch. And it is available now.

Why This Matters

If you have ever watched a chart and thought "that move came out of nowhere," it did not. There were signs in the order book — absorption at a key level, iceberg orders silently defending support, a cascade of liquidations waiting to trigger at a specific price, spoofed walls creating the illusion of resistance.

The problem was not that the information did not exist. The problem was that you did not have the tools to see it.

This update changes that. We are not talking about one widget or a minor enhancement. This is a complete market microstructure layer — the same class of tools that market makers, algorithmic trading firms, and prop desks rely on to see what everyone else misses.

What We Shipped

This was not a single feature drop. This was a ground-up construction of an entire order flow analysis layer — six new widgets, enhancements to four existing ones, three new exchange integrations, two market manipulation detection systems, and a full market replay engine.

Here is every piece.

1. Footprint Chart

The footprint chart is the single most important tool in an order flow trader's arsenal, and the hardest to build.

Every candle on a normal chart shows open, high, low, close. A footprint chart breaks each candle open and shows you the bid volume vs ask volume at every price level inside that bar. You see exactly where buyers and sellers fought, who won, and by how much.

What it detects automatically:

Imbalances — when one side overwhelms the other at a price level (3:1 ratio or greater), highlighted in colour so you spot them instantly
Stacked imbalances — three or more consecutive price levels all showing the same directional imbalance. This is one of the strongest signals in order flow analysis. When buyers dominate across a vertical stack, it often marks the beginning of a move.
Point of Control per bar — the price level with the highest total volume inside each candle
Unfinished auctions — when the high or low of a bar has zero volume on one side, meaning the auction at that extreme was never completed. Price tends to revisit these levels.
Net delta per bar — the total buy minus sell volume, rendered as a colour-coded bar so you see the character of each candle at a glance

This is the widget that platforms charge thousands for. We render it on HTML5 Canvas for performance, aggregate trades in real time from exchange APIs, and bucket them by price level per time interval. The engineering involved computing bid/ask classification for every trade, handling the precision differences across trading pairs, and rendering thousands of data points smoothly.

How to Read a Footprint Chart

If you are new to footprint charts, here is a practical walkthrough.

Imagine BTC/USDT prints a green candle from $64,000 to $64,500. A normal chart tells you "price went up." The footprint tells you everything that happened inside that move:

Price	Bids (Sells)	Asks (Buys)	Delta
64,500	12	45	+33
64,400	18	52	+34
64,300	85	22	-63
64,200	15	48	+33
64,100	20	41	+21
64,000	90	25	-65

At $64,000 and $64,300, sellers were dominant (high bid volume), but price still moved through those levels. That means buyers absorbed the selling and pushed through. Those become important levels — if price revisits them, there is a higher probability of support because the buyers who absorbed there are likely still positioned.

At $64,400 and $64,500, the buy imbalances are 3:1 or greater. These get highlighted automatically. Two consecutive imbalances means a stacked imbalance — an aggressive buyer is working a large order across multiple price levels.

The point of control is $64,300 (highest total volume: 85 + 22 = 107). This is where the most trading activity occurred, and it acts as a magnet for future price action.

The unfinished auction check looks at the top and bottom of the bar. If $64,500 (the high) shows zero bid volume, the auction at the top was never completed — sellers never tested that level. Price is statistically likely to revisit it.

This is the level of detail you get from every single bar on the chart, updating in real time.

2. Liquidation Map

Leveraged positions have mathematically deterministic liquidation prices. If you know the leverage and entry price, you know exactly where the exchange will force-close the position.

The Liquidation Map widget calculates predicted liquidation levels across leverage tiers from 2x to 125x, then overlays them as a heatmap against current price. You can see where the liquidation clusters sit — and when price approaches those zones, you know a cascade of forced selling or buying is about to hit.

What it shows:

Predicted liquidation price bands for longs and shorts across all leverage tiers
Real-time liquidation event markers from the exchange's forced order stream
A summary panel showing total long vs short liquidation volume and which side is more exposed
Squeeze direction indicator — which side will cascade first if price moves

This required integrating the futures API endpoints for mark prices, open interest, and the live forced liquidation WebSocket stream. The physics are straightforward but the data pipeline was not — liquidation events arrive in bursts during volatile moves and need to be aggregated into price zones in real time.

Why Liquidation Levels Are Tradeable

Liquidation cascades are some of the most violent moves in crypto. Here is how they work:

Price approaches a liquidation cluster (say, a concentration of long liquidations at $62,000)
Some positions begin to liquidate — the exchange sells their collateral at market price
Those forced sells push price further down
That triggers more liquidations at slightly higher prices
The cascade feeds on itself until the cluster is exhausted

The result is a sharp, sudden move that catches most traders off guard. But if you can see the cluster before price reaches it, you have context that others lack. You know that if $62,000 breaks, there is fuel for a rapid move to $61,200 where the next cluster sits. You can position accordingly — either by avoiding the trap or by trading the cascade itself.

The squeeze direction indicator tells you which side (longs or shorts) has more liquidation exposure near current price. When the answer is "shorts are concentrated 2% above" during an uptrend, you know there is fuel for a short squeeze if price pushes a little higher.

3. Multi-Exchange Aggregated Book

No single exchange shows you the full picture. A large order on one exchange might not exist on another. Arbitrage opportunities appear when prices diverge across venues. The true depth of a market is the sum of all venues, not any one of them.

We built data source integrations for three exchanges — with order book and trade data from each — and a real-time aggregation engine that merges them into a single unified view.

Three views:

Merged Book — all bids and asks combined, with colour-coded bars showing each exchange's contribution to every price level. You see the true aggregate depth.
Compare — side-by-side statistics per exchange: best bid/ask, spread, total depth. Instantly spot which venue has the tightest spread or deepest liquidity.
Arbitrage — when the best bid on one exchange exceeds the best ask on another, an arbitrage opportunity exists. The widget detects these automatically, shows the spread, and estimates profitability.

Exchange toggles let you include or exclude any venue. The backend fetches all three order books in parallel using async I/O, merges price levels (summing quantities where prices match across exchanges), and computes cross-exchange statistics in a single response.

The Fragmented Market Problem

Crypto is fundamentally different from traditional equities in one critical way: there is no consolidated tape. When you trade a stock, every order goes through a centralised matching engine or is reported to a regulatory body. When you trade crypto, there are dozens of independent order books with no obligation to synchronise.

This creates three problems that the Aggregated Book solves:

Hidden depth. A price level might show 5 BTC on one exchange but 50 BTC across all three. If you are only watching one venue, you massively underestimate the support or resistance at that level. The merged book reveals the true depth.

Best execution illusion. You might see a tight spread on your preferred exchange while a tighter spread exists elsewhere. The Compare view shows you which venue currently has the best liquidity conditions for your trade size.

Arbitrage visibility. When markets fragment, prices temporarily diverge. The Arb view catches these moments automatically. Even if you do not trade the arbitrage yourself, knowing that price is diverging across venues tells you something about the conviction behind a move. A rally confirmed across all three exchanges is stronger than one driven by a single venue.

4. Market Replay System

You cannot improve at reading order flow without reviewing what you missed. Professional traders study replays the way athletes study game film.

We built a complete market replay system:

Recording: Start a recording session for any symbol. The system captures full order book snapshots and trade data every two seconds, saved as compact JSONL files.
Playback: Load any recording and replay it at speeds from 0.5x to 10x. A transport bar with play, pause, stop, and skip controls. A mini order book and trade tape update in real time as the replay progresses.
File management: Browse all recordings, see file sizes and durations, delete old ones.

This means you can record a volatile session — an FOMC announcement, a liquidation cascade, an NFP release — and study the order book dynamics afterwards at your own pace. Pause on the exact moment a level broke, step through the absorption, see how the book reformed.

The backend records to disk as streaming JSONL (one JSON object per line per snapshot), which keeps memory usage flat regardless of recording length. Playback streams the file back with server-side pagination.

What to Record

The replay system is most valuable when you use it deliberately. Here are the sessions worth capturing:

High-impact news events. FOMC rate decisions, CPI releases, employment reports. The order book behaviour in the 30 seconds before and after is some of the most educational market data you will ever see. Watch how liquidity evaporates before the announcement (market makers pulling quotes) and how it snaps back with directional flow.
Liquidation cascades. When BTC drops 5% in 10 minutes, record it. In replay, you can slow it to 0.5x and watch the domino effect — how each liquidation level triggers the next, how the order book thins out during the cascade, and where it finally stabilises.
Your own trades. Record while you are actively trading. After the session, replay your entries and exits and see what the order book was telling you that you might have missed in real time.
Quiet accumulation periods. The boring sideways sessions where "nothing happens" often have the most interesting order book activity. Icebergs building, large orders being worked quietly, volume delta diverging from price. These are the setups that precede the next big move.

Start with recording 30-60 minutes of a volatile session. Replay it at 2x to get the overview, then drop to 0.5x for the key moments. You will learn more about market microstructure from one good replay than from a month of reading about it.

5. Iceberg Detection

An iceberg order is a large order that is hidden from the order book. The exchange only shows a small visible portion, and each time that portion is consumed, the exchange automatically refills it from the hidden reserve.

Icebergs are invisible on a normal order book display. But they leave a fingerprint: a price level that keeps getting consumed and keeps reappearing at the same quantity.

Our detection system maintains a rolling window of 30 order book snapshots (approximately 60 seconds of history). It compares consecutive snapshots and flags price levels where:

The quantity at that level was consumed (decreased significantly or disappeared)
The quantity reappeared at the same price within the next few snapshots
This refill pattern repeated three or more times

Confirmed icebergs are displayed in the Level 2 analytics tab with the price, side (bid/ask), refill count, and how long the iceberg has been active. When you see an iceberg on the bid side with 8 refills over 4 minutes, you know someone is defending that level with serious size.

Trading Around Icebergs

Icebergs are among the most reliable order flow signals because they represent committed capital. Someone is not just placing a limit order and hoping — they are actively working a large position and paying to keep it hidden.

Bid-side icebergs (support): A large hidden buyer is absorbing sell pressure at that price. The more refills, the more committed they are. If price is trending down and hits a level where an iceberg is detected with 5+ refills, that is a high-probability support zone. The hidden buyer has already absorbed significant volume and is still defending.

Ask-side icebergs (resistance): The mirror image — a large hidden seller is absorbing buy pressure. Multiple refills at resistance, especially during a rally, often mark a distribution zone where smart money is selling into strength.

The key insight is that icebergs show you where the conviction is. Anyone can place a visible limit order. Only someone with serious size and intent will run an iceberg algorithm that continuously refills after being consumed. When you see one, you are seeing institutional flow.

6. Spoofing Detection

Spoofing is the practice of placing large orders with no intention of filling them — the goal is to create false impressions of supply or demand to manipulate other traders. It is illegal in regulated markets, but it happens constantly in crypto.

Our detector tracks every large order that appears on the book (defined as 3x or greater than the average order size at that depth level). If a large order:

Appears on the book
Disappears within three snapshots (approximately 6 seconds)
And price never traded through that level (meaning it was pulled, not filled)

...it gets flagged as a potential spoof. The alert shows the price, quantity, lifetime, and timestamp. Over time, you develop a feel for which levels are genuine and which are being manipulated.

Spoofing in Practice

Spoofing is more common than most traders realise, especially in crypto markets where regulatory enforcement is limited. Here is what it looks like in real time:

You are watching BTC/USDT at $64,200. A massive sell wall appears at $64,500 — 150 BTC, far larger than any other level on the book. Retail traders see it and think "there is huge resistance at $64,500, I should sell." Price starts drifting down as people front-run the wall.

Then the wall vanishes. It was never real. The spoofer wanted to push price lower so they could accumulate at a better price. The 150 BTC order was a bluff, and everyone who sold into it just gave the spoofer cheaper coins.

Our detector catches this pattern. When you see a spoof alert, it means a large order just appeared and vanished without being filled. That information is actionable — it tells you the opposite of what the order implied. A spoofed sell wall that disappears is actually a bullish signal, because it means someone wanted to push price down (likely to buy).

The more spoofs you observe at a particular level, the more suspicious that level becomes. Real orders do not repeatedly appear and vanish. The spoofing alerts turn an invisible manipulation tactic into a visible signal.

7. Enhanced Existing Widgets

We did not just build new things — we went back and upgraded what was already there.

Volume Delta Widget — Added automatic divergence detection. The widget now compares CVD (cumulative volume delta) peaks and troughs against price peaks and troughs and flags bullish or bearish divergences with an explanation banner. Added a session reset button so you can zero out CVD at any point (useful for day session analysis).

Volume Profile Widget — Added a delta profile mode that splits each price level into buy volume and sell volume, rendered as stacked green and red bars. This shows you not just where volume occurred, but who was in control at each level. Added developing POC tracking that shows the Point of Control migration direction over time with trend indicators.

Order Flow Widget — Added order reconstruction. Exchanges fragment large orders into many small fills. A single 100 BTC order might appear as 47 separate trades. Our reconstruction algorithm groups fills by price, side, and time proximity (500ms window) back into their parent orders — revealing the true size of institutional trades that the raw tape hides. The widget shows a compression ratio so you can see how much was hidden.

Level 2 Widget — Now displays iceberg and spoofing alerts directly in the analytics tab alongside absorption detection, delta divergence, and the order book heatmap. The analytics tab has become a full microstructure command centre — every anomaly the system detects appears here with context.

Putting It All Together: A Real Trading Scenario

Here is how these tools work together in practice. This is not a hypothetical — it is the kind of setup that plays out multiple times per week on liquid crypto pairs.

Setup: BTC/USDT is trading at $65,200 after a 3% rally over the past 4 hours.

Liquidation Map shows: A dense cluster of short liquidations between $65,800 and $66,200. If price pushes another 1%, a cascade could trigger.

Footprint Chart shows: The last three 15-minute bars all have stacked buy imbalances between $65,000 and $65,200. Aggressive buyers are working the ask across multiple price levels. The delta per bar is increasingly positive.

Aggregated Book shows: All three exchanges confirm the buying. The merged book shows thin asks above $65,400 — not much resistance between here and the liquidation cluster.

Level 2 Analytics shows: An iceberg detected on the bid side at $65,100 with 6 refills over 3 minutes. A large buyer is defending the base of the move. No spoofing alerts on the ask side — the thin asks are genuine.

Volume Delta shows: CVD is making new highs in line with price — no divergence. The buying pressure is confirmed and sustainable.

Interpretation: This is a high-probability continuation setup. Aggressive buyers are pushing, a hidden buyer is defending the base, there is no sign of manipulation on the sell side, the asks above are thin, and a liquidation cascade sits 1% higher that would add fuel. A long entry with a stop below the iceberg level ($65,050) and a target at the liquidation cluster ($65,800) offers an asymmetric risk-reward.

Without order flow tools, all you see is "BTC went up 3%." With them, you see the full structure of why it went up and what happens next.

The Engineering

This was not a weekend project. The scope included:

6 new widgets with canvas-rendered visualizations
4 enhanced widgets with new analysis modes
3 exchange data integrations (all using public, unauthenticated endpoints for zero-friction setup)
5 new backend API route modules with parallel async data fetching
2 detection algorithms operating on rolling snapshot windows
1 recording/playback system with streaming file I/O
1 new preset dashboard layout specifically for order flow trading

Every widget follows the existing architecture: Zustand store integration, WebSocket for real-time updates, responsive grid layout, full help documentation, and consistent styling with the rest of the platform.

The multi-exchange aggregation engine uses asyncio.gather to fetch from all three exchanges in parallel, keeping latency to the slowest single response rather than the sum of all three.

The footprint chart aggregates thousands of individual trades per minute into price-level buckets with bid/ask classification, then renders them on Canvas with colour-coded imbalance detection — all updating in real time.

The iceberg and spoofing detectors run on a rolling window of order book snapshots. Every two seconds, the system takes a snapshot and compares it against the previous 30. This means the detection is continuous and stateless from the client's perspective — the backend does the heavy lifting of tracking which levels refilled, which large orders vanished, and whether price moved through them. The frontend just renders the results.

The market replay system was an interesting engineering challenge. Recording is straightforward — capture snapshots to disk as JSONL. But playback needs to feel like real-time data without actually streaming in real time. We solved this with server-side pagination: the client requests a time window, the server scans the JSONL file and returns the matching snapshots, and the client replays them locally at the chosen speed. This keeps the server stateless during playback while giving the client smooth transport controls.

The order reconstruction algorithm in the Order Flow widget uses a time-proximity clustering approach. Exchange APIs return individual fills, but a single institutional order might get matched against dozens of resting orders at different sub-price-levels within milliseconds. We group fills that share the same price, same side (buy/sell), and occur within a 500ms window into a single reconstructed order. The compression ratio — how many raw fills collapsed into how many reconstructed orders — tells you how fragmented the market is. A high compression ratio means large players are active and their true size is being hidden by the matching engine.

New Dashboard Layout: Order Flow

We added a dedicated Order Flow preset layout that puts all of these tools together in one screen:

Top row: Chart + Level 2 + Footprint Chart — the core microstructure view
Middle row: Volume Delta + Order Flow tape + Aggregated Book + Volume Profile — context and cross-venue analysis
Bottom row: Liquidation Map + Positions + Watchlist + Alerts — execution and risk

Switch to it from the layout menu. We also updated the Day Trader, Full Pro, and Crypto Focus layouts to incorporate the new widgets where they make sense.

Who This Is For

This suite is built for traders who want to understand why price moves, not just that it moved. If you have ever:

Wondered why price reversed at a seemingly random level (it was an iceberg)
Watched a breakout fail and retrace (spoofed orders created false momentum)
Seen a sharp move with no apparent catalyst (a liquidation cascade triggered at a cluster)
Wanted to see the "real" depth across multiple venues instead of trusting one exchange
Noticed that your chart-based entries keep getting stopped out by a few ticks before reversing in your direction (market manipulation at your stop level)

...this is your toolkit.

Which Tools to Start With

If this feels overwhelming, start small. You do not need to use all of these tools on day one.

Your Style	Start With	Add Later
Scalper (seconds to minutes)	Footprint Chart + Level 2	Order reconstruction, iceberg detection
Day trader (minutes to hours)	Volume Delta + Volume Profile (delta mode)	Footprint Chart, spoofing detection
Swing trader (hours to days)	Liquidation Map + Funding & OI	Aggregated Book, market replay
Researcher / student	Market Replay + any of the above	Everything — replay is your classroom

The Order Flow preset layout is designed for traders who want the full microstructure view from day one. The Day Trader and Crypto Focus layouts include the most relevant order flow tools without overwhelming you.

What This Replaces

If you are currently paying for dedicated order flow software, here is how our suite maps to features you might be using elsewhere:

Feature	Dedicated Platforms	Nydar
Footprint / cluster chart	$100-300/mo add-on	Included
Order book heatmap	$150-500/mo standalone	Included (L2 Heatmap tab)
Liquidation levels	$50-100/mo data subscription	Included
Multi-exchange aggregation	$200-500/mo professional	Included (3 exchanges)
Iceberg detection	Enterprise-only	Included
Spoofing alerts	Enterprise-only	Included
Market replay	$100-200/mo add-on	Included
Delta profile	$50-150/mo add-on	Included (Volume Profile widget)
Order reconstruction	Proprietary trading desks	Included (Order Flow widget)

We are not claiming feature parity with every institutional platform — some dedicated order flow tools have 20 years of development and dedicated hardware acceleration. But for the vast majority of order flow analysis that traders actually need, this suite covers it — and it is integrated directly into the same dashboard where you chart, screen, and manage positions.

Try It Now

Head to your dashboard, open the layout menu, and switch to the Order Flow preset. All widgets are available to add individually from the Add Widget menu under Market Data.

For detailed walkthroughs of each widget:

Footprint Chart Help
Liquidation Map Help
Aggregated Book Help
Market Replay Help
Level 2 / Order Book Help — includes iceberg and spoofing detection
Volume Delta Help — includes divergence detection
Volume Profile Help — includes delta profile mode
Order Flow Help — includes order reconstruction

DEV Community