→ About Obsidia
The honest strategy
marketplace for Solana traders.
Obsidia lets traders compose strategies with a no-code DSL, backtest them on real-fee OHLCV with no lookahead, and publish them to a marketplace where every alert and fill is auditable — explained by an AI co-pilot that shows its work.
Stage
Public beta
First market
Solana retail
Founder
Lopinti Abhijeeth Baba
Track
DeFi infra · Trader tooling
02 · The problem
The problem
The same overfit backtest, recycled across
every Solana trading channel.
TradingView
Strategies tested on closing prices, no fees, no slippage.
Telegram alpha
Bots delete losing calls. No proof of past performance.
Crypto X
Top traders ranked by retweets, not P&L.
Discord groups
Cherry-picked screenshots. No ledger.
Drift / Jupiter
No native way to track or verify a strategy's edge over time.
Onboarding
Each new tool re-asks for keys, channels, and trades.
6+
Disconnected tools every retail Solana trader stitches together — most asking for the same data.
0
Audited proof-of-edge between alpha groups. Every claim is a screenshot.
1
Retail trader holding the bag when the "200% backtest" goes -8% in production.
03 · Why it's a trust problem
Why this is a trust problem, not a UX problem
Public alpha, private numbers.
01
Backtests fudge fees and slippage
Most strategies look great on paper because the math skips fills. Add real Jupiter Perps fees and slippage and the edge often disappears.
02
No proof of past calls
Telegram bots, Discord channels, X traders — all delete or omit losing calls. No one keeps an honest ledger.
03
Ranking by attention, not performance
"Top trader" usually means most retweets. There is no neutral leaderboard backed by audited fills.
04
Retail absorbs the risk
The trader copying the call eats the fees, slippage, and downside. The publisher gets the upside in followers either way.
04 · Solution
Solution
One DSL. One audited backtester.
One alert pipeline that shows its work.
Honest backtester
Real fees, no lookahead
24 months of OHLCV at six timeframes, with Jupiter Perps and Drift fees baked in. Trade ledgers, equity curves, and drawdown — auditable down to the bar.
DSL marketplace
Composable strategies
A no-code DSL describes entry, exit, sizing, and fees. Authors publish; subscribers pay monthly for live signals — performance ranked by audited fills, not screenshots.
AI co-pilot
Every signal, explained
Each alert lands on Telegram, Discord, and email with a plain-English rationale: which signals fired, the multi-timeframe context, the historical pattern resolution.
05 · How it works
How it works
Compose. Backtest. Publish. Execute.
With proof at every step.
01 / Compose
No-code DSL builder
Stack entry conditions (RSI, BB-position, MTF alignment), choose exit type, sizing, leverage, venue. Save as a portable JSON spec.
Input · entry · exit · sizing · fees
02 / Backtest
Real fees, no lookahead
Run against 24 months of OHLCV with Jupiter / Drift fees and slippage. Inspect every fill, equity curve, and drawdown.
Output · trade ledger · equity · stats
03 / Publish
Marketplace listing
Optionally publish to the marketplace. Audited backtest, fee model, and live-trade history visible to every subscriber.
Pricing · monthly subscription
04 / Execute
Alerts → one-click trade
Live alerts on Telegram, Discord, email — each with AI rationale. Paper-trade or one-click execute on Drift / Jupiter Perps.
Output · live alerts · executed fills
Compounding
effect
effect
Every published strategy expands the canonical OHLCV + fee corpus. Every fill teaches the backtester. Every subscriber tightens the ranking — the marketplace gets harder to clone with every active week.
06 · Architecture
The deep-tech core
An umbrella DSL engine,
with vertical strategy agents that own their domain.
Layer 0 · Umbrella
DSL Backtester + Orchestrator
Owns the full pipeline — DSL parsing, OHLCV ingestion, no-lookahead simulation, fee + slippage application, persistence to Supabase, alert fan-out across channels.
A.01
Confluence
Multi-signal scoring (RSI, MACD, MTF, sweeps, fear/greed).
A.02
Mean Reversion
BB-band fades on Jupiter Perps with structured exits.
A.03
Realtime Spike
Tick-level scalp on sharp moves with stop and target rails.
A.04
AI Analyst
Order-flow + macro synthesis into a directional call with rationale.
A.05
Risk Sizing
Notional, % equity, ATR-risk; leverage bounded per plan.
A.06
Alert Router
Telegram + Discord + email fan-out with AI-generated rationale.
A.07
Strategy Auditor
Independent re-runs of subscribed strategies on stored OHLCV.
A.08
Memory
Per-user trade ledger, win/loss patterns, plan limits.
07 · Wedge
Beachhead
Solana retail traders on Drift + Jupiter.
One chain, deeply.
Why Solana first
- —The most liquid retail-perps surface in DeFi today, with strong native UX (Phantom, Jupiter, Drift).
- —Real fee + slippage data is reachable through public RPC and exchange WebSockets — no proprietary feed required.
- —Stablecoin rails (USDC, USDG) and SPL primitives are mature enough to support marketplace payouts.
- —Aligned with Solana Foundation's grant priorities — DeFi infra and trader tooling for the long-tail user.
Initial wedge specification
Geography
Solana, then Solana-adjacent EVM (Base, Arbitrum) via the same DSL
User
Retail / prosumer traders, $1k–$50k account size
Venues
Drift Protocol · Jupiter Perps · Hyperliquid (planned)
Buyer
Subscribers paying monthly per strategy; creators paid per subscriber
Pricing
Free · Pro $19 · Trader $49 · Quant $99/mo + creator payouts
08 · Moat
Defensibility
A frontend is easy to copy.
The execution layer underneath is not.
M.01
Real-fee OHLCV corpus
24 months × 6 timeframes × N venues, with fee + slippage models per venue — sharpens with every new strategy a creator publishes.
M.02
Composable DSL graph
Strategies are JSON specs. Reusable, forkable, comparable side-by-side — operational know-how as code, not a black box.
M.03
Audited backtest engine
Independent re-runs on stored OHLCV produce reproducible numbers. Creator can't quietly fudge results between subscriber updates.
M.04
Closed-loop signal data
Every alert + fill + subscriber outcome feeds the marketplace ranking. Generic LLMs can copy a UI but not the labelled outcome data.
M.05
Trust + audit surfaces
Consent for data sharing, transparent fee model, public trade ledger. A retrofit for ad-hoc Telegram alpha groups; native here.
09 · Stage
Where we are today
Public beta. Sharp thesis. Real trades on the books.
Current stage
Frontend live at obsidia.app on Vercel. Backend on Railway running 24/7 paper trading. 4 in-house strategies active, 155 closed trades on file, full Supabase persistence and live alerts via Telegram + Discord + email.
Initial validation
Founder lived experience
Three years building algo systems. Lost real money to overfit backtests — the precise pain Obsidia is built to remove.
Operating learning
155 paper trades across 4 strategy archetypes confirm the value of fee-honest backtests.
Adjacent paid demand
GMGN, Photon, BullX prove Solana traders pay for tooling. Backtest + audit is the unmet gap.
Sharper thesis
Repositioned from "honest backtester" to "strategy marketplace" — recurring revenue, network effects, defensible category.
M 1–2
Public launch, first 1k waitlist signups, design partner outreach.
M 3–4
Paid tier live. First 10 creator-published strategies on the marketplace.
M 5–6
Independent strategy auditor live. 50 creators, 500 subscribers.
M 7–8
One-click execute on Drift + Jupiter Perps. Live-fee data feed.
M 9–10
Multi-chain (Base, Arbitrum) via the same DSL. Pre-seed prep.