Daily Pace Estimation

Multi-Session Day Journey

Day 1 — Install Day

Key distinction: the daily pace bar's segments represent provider session windows — fixed 5-hour blocks anchored to the provider's reset schedule — not the user's working sessions. A user's continuous working period can span multiple provider windows, and a single provider window might contain work from different user "sessions" (e.g., morning and afternoon stints).

Pace math: this preview follows the CodexBar model from the refactor notes: compare actual minus expected in percentage points, not actual divided by expected. That keeps tiny daily windows stable: 1.1% against a 2.5% target is only 1.4 points in reserve, so it is still closest-to-target purple rather than ratio-based underuse.

In this example, Claude's provider session windows reset every 5 hours starting from a 2pm (14:00) anchor. The daily pace bar divides the day into: [00-04] [04-09] [09-14] [14-19] [19-24]. A developer installs Pitwall at 10am and starts working. They land mid-way through the 09:00–14:00 provider window. Pitwall has no history snapshots from before this window, so the estimation fallback activates for this segment.

At 2pm the provider session resets and the 14:00–19:00 window begins — but the user keeps working continuously. Their usage from 2pm–3pm now falls in this new provider window. Because Pitwall has been recording weekly snapshots since 10am, it has a real baseline from before 2pm and can compute an accurate weekly delta. The user eventually stops at 3pm, but the provider window runs until 7pm.

The user returns at 6pm for an evening stint. They're still in the same 14:00–19:00 provider window, so this usage accumulates in the same segment. When that window resets at 7pm and 19:00–24:00 begins, Pitwall again has recent snapshots as a baseline. The estimation is strictly a bridge for the first provider window after install — every subsequent window has real data.

Provider session windows (5h each, anchored to 14:00 reset)
00:00 – 04:00
04:00 – 09:00
09:00 – 14:00
14:00 – 19:00
19:00 – 24:00
User working sessions (when the developer is actually coding)
Working 10am–3pm
Working 6pm–9pm
00:0004:0008:0012:0016:0020:0024:00
User's 10am–2pm work → 09:00–14:00 provider window (estimated)
User's 2pm–3pm work → 14:00–19:00 provider window (weekly delta)
User's 6pm–7pm work → 14:00–19:00 provider window (same segment!)
User's 7pm–9pm work → 19:00–24:00 provider window (weekly delta)
Provider Window: 09:00–14:00
User starts working at 10am (mid-window)
Pitwall was just installed — no snapshots from before 09:00. Session util is 42% but daily pace shows ~1.1% (estimated). User works 10am–2pm in this window, then the provider session resets at 14:00.
Session 🚶
42%
Pace target this session: 40%; 2% ahead of pace.
Daily 🚶
1.1%
Daily target 12%; 1.1% observed in current window.
—/2.0%
—/2.5%
~1.1/2.5%
ESTIMATED
—/2.5%
—/2.5%
Weekly 🏃
52%
Pace target by now: 45%; 7% ahead of pace.
Estimation fallback (Daily only)
Provider Window: 14:00–19:00
Snapshots from 10am–2pm provide baseline
User's work from 2pm–3pm and 6pm–7pm both land here. Baseline: weekly was 52% at 1:55pm. Current weekly is 54.5%. Delta = 2.5% — accurate, covers both user stints.
Session 🛌
15%
New provider session started at 14:00. User resumed at 18:00.
Daily 🚶
2.5%
Daily target 12%; 2.5% observed in current window.
—/2.0%
—/2.5%
1.1/2.5%
2.5/2.5%
DELTA
—/2.5%
Weekly 🏃
54.5%
Pace target by now: 47%; 7.5% ahead of pace.
Weekly delta (authoritative)
Provider Window: 19:00–24:00
Snapshots from 14:00–19:00 provide baseline
User's 7pm–9pm work lands here. Baseline: weekly was 54.5% at 6:55pm. Current weekly is 55.5%. Delta = 1.0%. Earlier windows now show as pastObserved with real data.
Session 🛌
8%
New provider session started at 19:00. User working 19:00–21:00.
Daily 🦥
1.0%
Daily target 12%; 1% observed in current window.
—/2.0%
—/2.5%
1.1/2.5%
2.5/2.5%
1.0/2.5%
DELTA
Weekly 🏃
55.5%
Pace target by now: 48%; 7.5% ahead of pace.
Weekly delta (authoritative)

Day 2 — Morning-Only Session

Pitwall has been running since yesterday. At midnight the daily usage counter resets to 0% — the weekly counter (55.5% from Day 1) carries forward. Every provider window now has a clean baseline because Pitwall recorded snapshots throughout Day 1 and overnight. This means all segments use authoritative weekly deltas — no estimation fallback needed.

The user works only one stint today: 10am–2pm. This work spans the 09:00–14:00 provider window. After 2pm the user stops for the day. The remaining windows (14:00–19:00, 19:00–24:00) see zero new usage. Each completed window's actual contribution (including 0%) is compared against its expected pace share: daily target × (window hours / 24).

How unused windows affect daily pace: when the user doesn't code during a window, 0% is recorded against that window's pace expectation. The daily pace bar shows usage concentrated in the morning window (3.2% vs 2.5% expected — ahead of pace there) but the idle afternoon and evening windows drag the overall daily percentage down. By end of day the daily total is 3.2% of a 12% target — the user banked all their daily usage in one window.

Provider session windows (same 5h layout as Day 1)
00:00 – 04:00
04:00 – 09:00
09:00 – 14:00
14:00 – 19:00
19:00 – 24:00
User working sessions
Working 10am–2pm
00:0004:0008:0012:0016:0020:0024:00
Provider Window: 09:00–14:00
User works 10am–2pm, only active window today
Pitwall has overnight snapshots as baseline. Weekly was 55.5% at day start. User codes 10am–2pm, weekly rises to 58.7%. Delta = 3.2% — authoritative, attributed to this window.
Session 🦥
35%
Pace target this session: 40%; 5% behind session pace.
Daily 🦥
3.2%
Daily target 12%; 3.2% observed in this window (ahead of 2.5% window pace).
0/2.0%
0/2.5%
3.2/2.5%
DELTA
—/2.5%
—/2.5%
Weekly 🏃
58.7%
Pace target by now: 50%; 8.7% ahead of pace.
Weekly delta (authoritative)
Provider Window: 14:00–19:00
User stopped working — window is idle
User finished at 2pm. Provider session resets but no new usage arrives. Weekly stays at 58.7%. Delta = 0%. The window's 0% actual vs 2.5% expected pace means the daily rate is now falling behind the straight-line target.
Session 🛌
0%
No usage in this session window.
Daily 🛌
0%
Daily target 12%; 0% in this window vs 2.5% expected pace.
0/2.0%
0/2.5%
3.2/2.5%
0/2.5%
DELTA
—/2.5%
Weekly 🏃
58.7%
Pace target by now: 52%; 6.7% ahead of pace.
Weekly delta (authoritative)
Provider Window: 19:00–24:00
Day winding down — no further usage
Still no activity. Weekly remains 58.7%. The daily total is 3.2% from a single window vs the 12% target. Two consecutive idle windows (0/2.5% each) show how unused time accumulates as missed pace — the daily bar tells the story at a glance.
Session 🛌
0%
No usage in this session window.
Daily 🛌
0%
Daily target 12%; 0% in this window. Daily total: 3.2% of 12%.
0/2.0%
0/2.5%
3.2/2.5%
0/2.5%
0/2.5%
DELTA
Weekly 🏃
58.7%
Pace target by now: 53%; 5.7% ahead of pace.
Weekly delta (authoritative)

Day 3 — Heavy Usage Day

The user has a busy coding day with three working stints: 10am–2pm, 3pm–7pm, and 8pm–11pm. Each stint generates usage in its respective provider window. All segments remain authoritative deltas — Pitwall has continuous snapshot history from Day 1 and Day 2.

How heavy usage distributes across windows: unlike Day 2 where usage was concentrated in one window, today's usage spreads across three windows. Each window independently compares actual vs expected pace (2.5% per 5h window). A window where actual exceeds its pace share shows the user is consuming faster than the daily straight-line rate during that period.

Daily cumulative view: by the end of the 19:00–24:00 window, the daily bar shows 4.0 + 3.5 + 2.8 = 10.3% against a 12% daily target. The sleeping windows (00–04, 04–09) contributed 0%, but the three active windows all exceeded their 2.5% pace share, keeping the user close to target despite not using the full 24 hours. Weekly rises from 58.7% to 69.0%.

Provider session windows
00:00 – 04:00
04:00 – 09:00
09:00 – 14:00
14:00 – 19:00
19:00 – 24:00
User working sessions
Working 10am–2pm
Working 3pm–7pm
Working 8pm–11pm
00:0004:0008:0012:0016:0020:0024:00
Provider Window: 09:00–14:00
Heavy morning session: 10am–2pm
Weekly was 58.7% at day start. User codes intensively 10am–2pm, weekly rises to 62.7%. Delta = 4.0% — well above the 2.5% window pace. This single window already covers a third of the 12% daily target.
Session 🔥
65%
Pace target this session: 40%; 25% ahead of session pace.
Daily 🦥
4.0%
Daily target 12%; 4.0% in this window vs 2.5% expected pace.
0/2.0%
0/2.5%
4.0/2.5%
DELTA
—/2.5%
—/2.5%
Weekly 🏃
62.7%
Pace target by now: 55%; 7.7% ahead of pace.
Weekly delta (authoritative)
Provider Window: 14:00–19:00
Afternoon session: 3pm–7pm
User returns at 3pm after a break. Works through the full provider window until 7pm. Weekly rises from 62.7% to 66.2%. Delta = 3.5%, again exceeding the 2.5% window pace. The 09–14 window now shows as past with its confirmed 4.0% contribution.
Session 🔥
70%
Pace target this session: 40%; 30% ahead of session pace.
Daily 🚶
3.5%
Daily target 12%; 3.5% in this window vs 2.5% expected pace. Running total: 7.5%.
0/2.0%
0/2.5%
4.0/2.5%
3.5/2.5%
DELTA
—/2.5%
Weekly 🏃
66.2%
Pace target by now: 57%; 9.2% ahead of pace.
Weekly delta (authoritative)
Provider Window: 19:00–24:00
Evening session: 8pm–11pm
User's third stint of the day. Weekly rises from 66.2% to 69.0%. Delta = 2.8%, still above the 2.5% window pace. All five daily segments now have confirmed values — the complete daily picture shows 10.3% of the 12% target consumed across three active windows and two sleeping windows.
Session 🔥
55%
Pace target this session: 40%; 15% ahead of session pace.
Daily 🚶
2.8%
Daily target 12%; 2.8% in this window vs 2.5% expected pace. Daily total: 10.3%.
0/2.0%
0/2.5%
4.0/2.5%
3.5/2.5%
2.8/2.5%
DELTA
Weekly 🏃
69.0%
Pace target by now: 58%; 11% ahead of pace.
Weekly delta (authoritative)