TradesViz Feature Review: Evaluating the New Options Backtesting and Exit Diagnostics

TradesViz shipped a five-month feature drop covering April through August 2026, and the changelog reads like the standard retail-analytics bingo card: an AI trading coach, Plaid-based broker sync, mobile v3, options backtesting, plus a new exit-diagnostics module. I pulled the announcement, filtered out the marketing noise, and looked for the one or two pieces that actually move the needle for systematic traders. Most of it is decoration. One part deserves a real test.

What actually shipped

The load-bearing feature is the historical options backtesting engine. Most journaling platforms treat multi-leg structures as a CSV footnote — Greeks approximated badly, fills optimistic, chain data spotty. If TradesViz can re-run historical SPX and equity option spreads with realistic slippage assumptions and surface accurate Greeks in the post-trade report, it closes a real workflow gap. I want to see the fill model before I trust any number it outputs.

Plaid-based sync across third-party brokers is table stakes in 2026, but it matters. Manual imports create dataset gaps, and gaps quietly invalidate backtests — you don't know your edge is fictional until a strategy blows up against a survivorship-clean dataset. Mobile v3 is cosmetic until the session replay works at full tick granularity on a phone. That would let me audit MFE/MAE distributions on the commute instead of at the desk.

The exit diagnostic piece is the one to watch

"Exit Insights" is the most interesting item if it does what the name suggests: surfaces slippage, MFE/MAE per trade, and expectancy broken down by exit reason. That audit is the closest thing to a free edge retail traders get, because almost nobody actually runs it. An exit-reason tag on every trade, combined with realized R-multiple distribution, exposes whether your "discretionary" cut is mathematically better or worse than your stop. If the module just tells you "you exited too early," skip it. That is a narrative, not a metric.

Where I'm skeptical

The AI trading coach is where I stop being polite. Every platform now ships an LLM that summarizes your journal and calls it coaching. Pattern matching on a single trader's own entries is not edge — the sample size is absurdly small, and the bot will confidently narrate noise as signal. Worse, traders tend to substitute the bot's story for an actual statistical review. It goes nowhere near any live decision I make. The whole "coach" category is a chatbot dressed in trading vocabulary.

How I'll stress it

I'm running the options engine against my own historical SPX spread log from Q1–Q2 2026. If the reported Greeks match what I actually booked, and the slippage assumption is honest rather than rosy, the tool earns a permanent seat in the workflow. If not, it joins the pile of dashboards that look sharp on screenshot day and useless on data day. Plaudits to any retail platform that ships a backtester with deliberate pessimism baked into its fill model — that is rarer than it should be.

For traders weighing analytics tooling against execution stack — sync plumbing matters as much as the dashboard it feeds, and the Australian share-trading platforms comparison is a useful parallel on how broker connectivity shapes the data you can even analyze in the first place.