Cryptomaty, Crypto Strategy Platform

Crypto Strategy Platform Case Study

I made an AI coin screener, no-code strategy templates and backtesting read as one workflow, for 19,000+ traders who had never written a line of code.

Scope of workWeb and Mobile Crypto Product

Timeline8 weeks

RoleProduct Designer

Team1 PM, 2 Engineers

Visit Cryptomaty View all projects

About Cryptomaty

Cryptomaty is an automated crypto trading platform used by more than 19,000 traders. An AI screener flags coins turning bullish or bearish, traders pick or build a no-code strategy, backtest it against historical market data, connect an exchange account and deploy it in one click, then track P&L per strategy and per broker. This case study covers how that screen, backtest and deploy path was redesigned to stay readable for first-time and experienced traders at once.

cryptomaty.com

Cryptomaty: Crypto Strategy Platform Case Study overview

The problem

Cryptomaty's AI screener, strategy templates and backtest engine were each technically strong, but a new trader could not tell a conservative template from an aggressive one, judge why the screener had called a coin bullish, or read what a backtest result was actually saying about risk.

User pain

New traders could not tell a conservative template from an aggressive one.
The AI screener called coins bullish or bearish without showing its working.
Backtest output was data, not a go/no-go read on the strategy.

Business pain

Traders browsed templates but stalled before connecting an exchange.
A free-for-everyone plan brought in beginners the interface was not written for.
Support absorbed questions about what a deployed strategy was currently doing.

Product gap

Screener, templates, backtest and live P&L behaved as four separate tools.
Risk level was buried in strategy parameters instead of shown on the template.
Nothing carried a backtest result forward into the decision to deploy.

What I owned

Owned

AI screener listing and the reasoning panel behind each verdict
Strategy template cards, including the risk tier on the card face
Backtest report and the go / no-go read that closes it
Deployed-strategy P&L, split per strategy and per broker

Collaborated

Engineering on exchange API connection states and failure modes
PM on what the free tier exposes and where it stops
Data on which screener signals were stable enough to explain

Out of scope

The screener's model and its signal generation
Exchange integrations and custody
Growth, referrals, and the points programme

The results

35%

faster strategy backtest setup

+26%

deploy-ready workflow completion

-31%

support confusion on strategy states

Traders moved from screener to live strategy without losing the thread. Screen, build, backtest and deploy became one measurable path. A free tier could take in beginners without buying support load.

The approach

Two audiences, one surface. I mapped the points where a first-time and an experienced crypto trader diverge, then designed the shared path so it stays legible to both rather than optimal for neither.

Watched first sessions

Followed new traders from sign-up to the point they stalled, almost always the template grid.

Made risk visible

Pulled risk tier out of the strategy parameters and onto the template card face.

Gave the screener a why

Attached the three reasons behind each bullish call, so a verdict could be checked instead of trusted.

Closed the backtest

Turned the report into one sentence and two doors.

Before the pixels

The boxes-and-arrows stage. Each of these settled a question the finished screens no longer show you was ever open.

The screener is a starting point, not a verdict

A bare verdict asks to be trusted. A verdict carrying its three reasons asks to be checked, and that is the one traders actually acted on.

Which risk level to deploy at

Same grid, one chip added. Moving the tier onto the card face turned 'which strategy do I run' into 'what risk am I willing to deploy at', a question a first-time trader can actually answer.

Backtest read as a go / no-go

Every number in this sketch was already on the old report. None of them said go or no, so the read got written, and the two doors sit under it.

The decisions

The calls that shaped the product, with what each one cost. Anything crossed out here was a real option at the time.

Decision 1: Template onboarding

Traders opened expert-built templates with no read on whether one suited them.

Considered

Template gallery only
Risk-tiered onboarding
Expert mode default

Chose

Risk-tiered onboarding with strategy fit guidance

Gave up

One extra instructional layer

On what evidence

Drop-off concentrated immediately after template selection

What happened

Backtest setup time reduced by 35%

Decision 2: Backtest story

Backtest reports were thorough and left the quality judgement to the trader.

Considered

Raw metrics table
Outcome scorecards
Chart-only summaries

Chose

Outcome scorecards with plain-language highlights

Gave up

More curated content to maintain

On what evidence

Traders asked for a direct read, not more charts

What happened

Higher deploy-readiness completion

Cryptomaty: options considered for Decision 2: Backtest story

Decision 3: Deploy handoff

Backtesting and one-click live deployment felt like separate products sharing a dashboard.

Considered

Direct deploy button
Checklist gate
Modal warning

Chose

Checklist gate tied to strategy confidence indicators

Gave up

Added one confirmation step

On what evidence

Support logs showed confusion about what a live strategy was doing

What happened

State confusion tickets reduced by 31%

The work

Risk level became a first-class attribute on every template rather than a number buried in the parameters, and backtest output was reframed into a go/no-go read with the supporting detail available on demand.