FXP vs XPP: which held to its multiple?

Over three months against its own daily promise, FXP finished 2.8 points over and XPP 1.5 points short. ProShares UltraShort FTSE China 50 and ProShares Ultra FTSE China 50.

−10.8%
FXP returned, 3 months
+10.9%
XPP returned, 3 months
+2.0 pts
FXP from its stated multiple
−1.9 pts
XPP from its stated multiple
FXP · 3 months to Sep 30, 20262 pts ahead of its stated multiple
2 pts ahead of its stated multipleFXP returned −10.8% while −2 times FXI's move would have been −12.8%FXI +6.4% ×−2 implies−12.8%FXP returned−10.8%2 pts ahead of its stated multipleFXP returned −10.8% while −2 times FXI's move would have been −12.8%FXI +6.4% ×−2 implies−12.8%FXP returned−10.8%

FXP returned −10.8% while −2 times FXI's move would have been −12.8%

XPP · 3 months to Sep 30, 20261.9 pts short of its stated multiple
1.9 pts short of its stated multipleXPP returned +10.9% while 2 times FXI's move would have been +12.8%FXI +6.4% ×2 implies+12.8%XPP returned+10.9%1.9 pts short of its stated multipleXPP returned +10.9% while 2 times FXI's move would have been +12.8%FXI +6.4% ×2 implies+12.8%XPP returned+10.9%

XPP returned +10.9% while 2 times FXI's move would have been +12.8%

ETFIQ Decay Resistance Score · XPP scores higherDid it keep up with its own daily multiple, compounded day by day?

FXP among the 125 inverse ETFs over three months

FXP 79.6
0.4, the lowest in this set99.6, the highest

XPP among the 470 leveraged ETFs, long, over three months

XPP 86.8
0.1, the lowest in this set99.9, the highest

A percentile among the 125 inverse ETFs over three months. XPP is a percentile among the 470 leveraged ETFs, long, over three months, a different set, so the two marks are not on one scale. It is a position in a set, not a rating, and neither end of it is a recommendation. All leveraged ETFs ranked by it → How it is computed →

Performance, window by window

Total returnMultiple would giveDifference
WindowFXPXPPFXPXPPFXPXPP
1 month+8.8%−8.2%+7.6%−7.6%+1.2 pts−0.5 pts
3 months−10.8%+10.9%−12.8%+12.8%+2.0 pts−1.9 pts
6 months+7.1%−11.2%+7.2%−7.2%−0.1 pts−4.0 pts
1 year+37.1%−35.7%+31.5%−31.5%+5.6 pts−4.1 pts
3 years−64.1%+24.4%−76.7%+76.7%+12.6 pts−52.3 pts
Since launch
FXP Jan 2010 · XPP Jan 2010
−98.0%−69.8%−33.8%+33.8%−64.2 pts−103.6 pts

FXP and XPP over each window. Every figure is an ETFIQ calculation with distributions reinvested. Source: ETFIQ. Open the live comparison on ETFIQ →

On the same fields

FXP
ProShares UltraShort FTSE China 50 · Aims to return twice the opposite of the daily move of iShares China Large-Cap ETF (FXI)
XPP
ProShares Ultra FTSE China 50 · Aims to return twice the daily move of iShares China Large-Cap ETF (FXI)
Issuer ProShares ProShares
Sets out to return -2x +2x
Underlying asset FXI FXI
Segment country country
Fund returned, 3 months or since launch −10.8% +10.9%
Underlying returned, over that window +6.4% +6.4%
What the stated multiple implies, over that window −12.8% +12.8%
Difference from stated, over that window +2.0 pts −1.9 pts
Fund returned, 1 year or since launch +37.1% −35.7%
Difference from stated, over that window +5.6 pts −4.1 pts
Underlying volatility 17% 17%
Difference over the days both have traded no shared window −1.9 pts
Expense ratio 0.95% 0.95%
Launched Jan 4, 2010 Jan 4, 2010
Net assets $6m $8m

FXP and XPP on the same fields, as of Sep 30, 2026. Source: ETFIQ.

FXP in plain words

Three months to Sep 30, 2026: FXP returned −10.8% where its own daily promise gave −13.7%, 2.8 points over. Read the multiple against the whole window instead and −2 times FXI's 6.4% implies −12.8%, which makes FXP look 2.0 points over. 0.8 of that is daily compounding, which happens to any −2 times fund over the same path, and the rest is the fund. FXP aims to return -2 times FXI's move each day, then resets. Over one day it does that. Over longer, the daily results compound, so the total is not -2 times the period's move: in a market that falls and comes back it is reliably less, and in a steady run it can be more. FXI moved at 17% annualized over that window. That is what decides how far the two figures separate: the same fund on a calm underlying and a violent one is a different product.

XPP in plain words

Three months to Sep 30, 2026: XPP returned +10.9% where its own daily promise gave +12.4%, 1.5 points short. Read the multiple against the whole window instead and 2 times FXI's 6.4% implies +12.8%, which makes XPP look 1.9 points short. 0.4 of that is daily compounding, which happens to any 2 times fund over the same path, and the rest is the fund. XPP aims to return +2 times FXI's move each day, then resets. Over longer, the daily results compound, so the total is not +2 times the period's move: in a market that falls and comes back it is reliably less, and in a steady run it can be more.

Questions people ask

Which came closer to its stated multiple, FXP or XPP?
Over the window to Sep 30, 2026, FXP finished 2.0 points from what its multiple implies and XPP finished 1.9 points from its own, so XPP came closer. Neither figure predicts the next window: it depends on how much the underlying moves about.
Are FXP and XPP levered on the same thing?
Yes. Both are levered on iShares China Large-Cap ETF, FXP at -2 times and XPP at +2 times the daily move.
Which one decays faster, FXP or XPP?
Decay follows how much the underlying moves about. Over this window FXP’s moved at 17% annualized and XPP’s at 17%, so FXP has the rougher ride and, at the same multiple, loses more to compounding.
Can I hold FXP or XPP for a year?
Both reset every day, so the multiple in the name applies to a single day and the daily results compound. Over a year the total is not the multiple times the year’s move, and in a market that falls and comes back it is reliably less. ETFIQ makes no recommendation either way.
Which is cheaper, FXP or XPP?
FXP charges 0.95% a year and XPP charges 0.95%, so FXP is cheaper. Fees come from each fund's prospectus.
Cite this page

ETFIQ, FXP against XPP, data as of Sep 30, 2026. https://etfiq.com/compare/leverage/fxp-vs-xpp

Open data

Free to use with attribution. Every figure is calculated from a named public source; the method is at etfiq.com/methodology.