Week in Review: Apr 7 - Apr 13, 2026
One of the more productive weeks in a while. Multiple threads all pushed forward significantly — new territory with algorithmic trading, a big sprint across the app portfolio, and two flagship apps shipped to beta testers with meaningful feature upgrades. Here’s what actually happened.
Algorithmic Paper Trading
I built a paper trading system from scratch this week. The goal is straightforward: test multiple trading strategies in isolation, compare performance over time, and eventually decide if any of them are worth running with real money. Paper trading means no actual funds are at risk — it’s running against a simulated account with real market data.
The system ended up with six strategies, each running in a completely isolated account so there’s no cross-contamination between results. On the stock side: VWAP Reversion (trade when price deviates from volume-weighted average), Opening Range Breakout (trade the direction of the first 30 minutes of market action), EMA Trend (follow the trend using exponential moving averages), RSI Reversion (mean reversion when overbought or oversold), and Gap & Go (trade gaps from the previous close). On the crypto side: a momentum strategy that runs 24/7 since crypto doesn’t close.
All orders are limit orders — no market orders — with 20% max position sizing and a 5% daily loss halt that stops trading if things go badly enough. I wrote a dashboard that aggregates all six accounts in real time, so I can see P&L and open positions at a glance.
The first week of market data comes in next week. The real question isn’t “does this work in theory” — there’s research for that — but which specific parameters and risk settings perform best in current market conditions.
App Sprint
The app portfolio has had a backlog of APKs waiting to be built. This week I knocked out 13 of them in a sequential build pipeline: ZenTick, Metronome, SaleFinder, UnitConverter, FlipRoll, Flashcards, PriceCompare, LoanCalc, QuoteVault, SleepSounds, ColorPicker, WarrantyTracker, and GarageKeep. All bumped to v0.2.0 with signed release APKs.
In parallel, I enhanced another 13 existing apps with visual improvements — progress bars, chart components, and better stat displays that make the data actually readable at a glance. Apps that showed raw numbers now show visual representations of the same data.
Getting APKs built isn’t the hard part. The hard part is everything that needs to happen before they hit a store: store listings, icons, feature graphics, privacy policies, content ratings. That pipeline is now fully automated for 39 apps. Titles, descriptions, icons, and feature graphics are all generated and ready. The Tier 2 apps (16 of them) have AABs built and a submission script ready to run. The only manual step left is creating the app listings in Play Console — I need to do that in the browser, about 2-3 minutes per app, and then workers can take over and submit everything automatically.
It’s the last manual gate before 37 apps go live. I need to carve out a few hours and just do it.
RxLog v1.2.4
Four releases shipped this week for RxLog, ending at v1.2.4. The most meaningful additions in this cycle:
Symptom editing. Previously, logging a symptom was a one-way operation — you could add entries but not fix mistakes. Now symptom entries are expandable and tappable with an Edit button that reopens the original form pre-populated. Small quality-of-life fix that eliminates a lot of frustration.
Doctor Notes screen. This ended up being a more useful feature than I expected when I started it. You can build a list of topics you want to bring up at your next appointment, attach each one to a specific doctor, and check them off after the visit. Pending items sort to the top, resolved items move to a “discussed” section below. It turns the app into a light medical diary and appointment prep tool rather than just a medication tracker.
IAP fixes. Some edge cases in the in-app purchase flow for Pro were causing silent failures on certain Android versions. Fixed the SKU handling and offer token logic that was tripping on v12+ of the purchase library.
Beta tester feedback has been positive on both new features.
SiteSnap v0.10.2
SiteSnap went through several releases this week too, landing at v0.10.2.
The biggest addition was a full Pro unlock system via in-app purchase (v0.10.0). The free version keeps all core functionality; Pro removes ads and unlocks higher-resolution PDF exports, advanced annotation tools, and weather data on reports. Building IAP into an Expo app is messier than it should be — there are native Kotlin patches required, mismatched dimension strategy flags in Gradle, and a handful of library quirks that need specific workarounds. Got it all sorted and working.
Other additions in the v0.10.x cycle: dashed and dotted pen styles for annotations, logo placement controls for PDF exports (left / center / right), custom checklist templates that persist across sessions, and a fix for photos-per-page controls that were showing even when the selected template doesn’t support them.
Home Assistant Repair
The HA configuration had accumulated 48 repair issues over time. Worked through 46 of them this week:
Migrated a batch of legacy platform: template sensor configs to the modern template: block format that HA now expects. This is a deprecation that’s been in the warning state for a while — HA keeps it working but flags it in the repair dashboard. Updated four automations that were still calling notify.charles instead of notify.mobile_app_charles_phone, which HA renamed at some point and stopped recognizing. Removed a stale ping sensor block in the dashcam package that referenced a deprecated YAML format.
Two items remain: one orphaned Keymaster lock entry that needs to be deleted through the HA UI directly, and a Samsung TV integration that needs a manual reauthentication. Both require clicking through the UI — can’t script them.
Next Up
Trading bot results start coming in next week. More interesting after a few weeks of data, but I’ll be watching the first trades.
The Play Console listing session is the highest-priority thing I’ve been procrastinating. A few hours of browser work unlocks 37 apps for submission — no code needed, just clicking through Google’s forms. On the list for this week.
RxLog has more feedback to process. Nothing urgent — mostly polish and edge cases.
Enjoyed this post?
Subscribe to get notified when I publish new articles about homelabs, automation, and development.
No spam, unsubscribe anytime. Typically 2-4 emails per month.