Good morning · overview
Strategies
Needs a look
What happened
Strategy mixer · tick strategies to stack them and see the total
Period:
Costs:
Rebalance:
Portfolio value, USD · log scale
What could go wrong · the next 12 months
Performance
By strategy · live vs. what the backtest expects
Cumulative P&L by strategy, USDC (closed trades)
Execution quality · what trading really costs vs. the backtest
Costs per trade
From the exchange's own fill records (every fill's fee and maker/taker flag, plus funding), copied read-only every 15 minutes.
Maker = the order rested on the book (Bybit 0.02%); taker = it crossed the spread (0.055%). Entries try maker for 20 s first, signal exits too
(stops always fill as taker). "Backtest" is the fee per round trip the backtest and the review assume. Funding: + = paid, − = received.
What changed, and when (UTC)
Every strategy on every account
Cumulative P&L, all accounts together (USD, closed trades)
Every closed trade
Growth of $1 · log scale
Return by calendar year
A walk-forward re-picks each strategy's settings using only the past (the training window), then trades the next window with them. Stitching the out-of-sample windows together gives the result you could actually have had without hindsight.
Astro X watcher · @astronomer_zero
Copy trades (closed)
Every original post is read by the LLM; the badge shows what the copy sleeve did with it. Only BTC calls with a target are traded (1% stop, all out at his first target, 14 days max). Click a post to open it on X.
Crowd sentiment · Astro's checklist
Context, not a signal: since 2020 these readings did not time turns on their own (extreme fear was usually
followed by more downside, greed by more upside - data/reports/sentiment.txt). Astro uses them as a precondition, then waits for confirmation.
MEMORY SA · paper stat arb (SNDK/MU · SNDK/DRAM)
z-score, last 14 days (enter at ±2, exit at 0, stop at ±4)
Paper trades
Forward test of the backtest (Reports → stock_pairs). Hedge ratio and z-score bands refit monthly from the last 60 US trading days
of the real stocks; live prices and funding from the Lighter perps; signals only during US market hours. $10,000 paper capital, one slot per pair,
about 1x per leg, 3 bps per fill. Every open/close goes to Telegram (#memory_pairs) with leg sizes, so it can be followed by hand. No real orders.
On-chain volume · Solana · Base · BNB Chain · Robinhood Chain
Daily DEX volume, 7-day average ($m)
Source: DefiLlama (daily DEX volume). Heat = 7-day average ÷ 90-day average: ≥ 1.5x = heating up
(Telegram ping - the cue to revisit the parked Meteora LP strategy), ≤ 0.7x = cooling. Refreshed every 6 hours.
Funding arb scanner · Bybit · Hyperliquid · Aster · Variational · Lighter · Extended · trade.xyz
Margin plan · from your wallet to each venue (manual transfers until the real-money stage)
Paper book · opened automatically on each alert, accrues the real funding that follows
Read-only: no orders, no transfers. Net return = funding gap minus round-trip fees and spreads on both legs, spread over
the expected hold (Settings → Funding arb scanner). Coins are matched by ticker across venues (GOLD = XAU, CL = WTI, ...); a pair only counts when both
legs' prices agree, which drops same-ticker clashes (a stock vs. a token) - still check the coin before acting. Extended and trade.xyz
list stocks and commodities: their funding can behave differently around market closes and weekends. Funding above ~±500%/yr is usually a short squeeze that fades within hours.
Paper bot –
Stop blocks new entries; an open position still exits on its stop or an opposite cross. Fills use the live Variational spread.
Manual mode (Omni) –
The bot never trades your wallet. It issues tickets; you place the order in Omni and confirm the fill price here.
Last signal bar
Pending tickets
No pending tickets.
Paper account
Manual account (your Omni position)
ETHUSDT · EMA · VWAP
Changes apply to the running bot within a few seconds (● = differs from default). Changing the timeframe or strategy while a position is open keeps that position and its stop.
Period:
Equity (USDC)
Drawdown %
By year
Long vs short · exits
Monte Carlo (1,000 reshuffles, liquidation from intrabar MAE)
Scenario sweep
Trades
Saved research reports
Run your own optimizer
Runs every combination in the grid over the full history (at 1x, with Variational costs), then a walk-forward test: tune on a train window, trade the next unseen test window, repeat. The walk-forward result is the honest estimate; the full-history top list is in-sample and optimistic.
Walk-forward windows (out-of-sample)
Top 25 over full history (in-sample)
Full bid/ask spread in bps from Variational's public stats API (polled every 60s; Variational may cache quotes up to 600s). Leave the worker running for 2–4 weeks, then update the spread table in Settings.
Upload the Strategy Tester "List of Trades" CSV from pine/vwap_ema9_variational.pine to check that the bot reproduces TradingView's entries. Live alerts arrive at /api/tv/webhook and are compared bar by bar.
Recent TradingView alerts
Strategy sleeves trading real orders on the Hyperliquid testnet (fake money) through python -m vsbot.live. Stops rest on the exchange. Kill switch, daily loss limit and position cap are in Settings → Live.
Runner –
State –
Equity –
Peak / drawdown –
Today –